About the training process

Hello! I run the training code with the default config file you provided on the full training dataset . But when I run the test code on the 3DPW dataset, I got the different results when PA-MPJPE is 57.2mm while result in the paper is 51.9mm. Do I miss something?

Hi @ZephirGe,

I can reproduce similar results. Could you share your training log? That should be saved as a txt file under the output directory.

2020-04-28 11:32:21,737 GPU name -> Tesla V100-SXM2-32GB 2020-04-28 11:32:21,737 GPU feat -> _CudaDeviceProperties(name='Tesla V100-SXM2-32GB', major=7, minor=0, total_memory=32480MB, multi_processor_count=80) 2020-04-28 11:32:21,738 {'CUDNN': CfgNode({'BENCHMARK': True, 'DETERMINISTIC': False, 'ENABLED': True}), 'DATASET': CfgNode({'SEQLEN': 16, 'OVERLAP': 0.5}), 'DEBUG': False, 'DEBUG_FREQ': 5, 'DEVICE': 'cuda', 'EXP_NAME': 'vibe_raw', 'LOGDIR': 'results/vibe_raw/28-04-2020_11-32-21_vibe_raw', 'LOSS': {'D_MOTION_LOSS_W': 0.5, 'KP_2D_W': 300.0, 'KP_3D_W': 300.0, 'POSE_W': 60.0, 'SHAPE_W': 0.06}, 'MODEL': {'TEMPORAL_TYPE': 'gru', 'TGRU': {'ADD_LINEAR': True, 'BIDIRECTIONAL': False, 'HIDDEN_SIZE': 1024, 'NUM_LAYERS': 2, 'RESIDUAL': True}}, 'NUM_WORKERS': 8, 'OUTPUT_DIR': 'results/vibe_raw', 'SEED_VALUE': -1, 'TRAIN': {'BATCH_SIZE': 32, 'DATASETS_2D': ['Insta', 'PoseTrack', 'PennAction'], 'DATASETS_3D': ['ThreeDPW', 'MPII3D'], 'DATASET_EVAL': 'ThreeDPW', 'DATA_2D_RATIO': 0.6, 'END_EPOCH': 30, 'GEN_LR': 5e-05, 'GEN_MOMENTUM': 0.9, 'GEN_OPTIM': 'Adam', 'GEN_WD': 0.0, 'LR_PATIENCE': 5, 'MOT_DISCR': {'ATT': {'DROPOUT': 0.2, 'LAYERS': 3, 'SIZE': 1024}, 'FEATURE_POOL': 'attention', 'HIDDEN_SIZE': 1024, 'LR': 0.0001, 'MOMENTUM': 0.9, 'NUM_LAYERS': 2, 'OPTIM': 'Adam', 'UPDATE_STEPS': 1, 'WD': 0.0001}, 'NUM_ITERS_PER_EPOCH': 500, 'PRETRAINED': '', 'PRETRAINED_REGRESSOR': 'data/vibe_data/spin_model_checkpoint.pth.tar', 'RESUME': '', 'START_EPOCH': 0}} 2020-04-28 11:33:06,716 => no checkpoint found at '' 2020-04-28 11:33:53,067 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 5.2513 | loss_kp_2d: 2.22 | loss_kp_3d: 0.71 | e_m_disc_loss: 0.22 | d_m_disc_real: 0.10 | d_m_disc_fake: 0.09 | d_m_disc_loss: 0.19 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:34:05,446 (67/67) | batch: 117.3ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 11:34:38,198 Epoch 0, MPJPE: 91.4964, PA-MPJPE: 55.7500, ACCEL: 28.5521, PVE: 108.7415, ACCEL_ERR: 29.3455, 2020-04-28 11:34:38,326 Epoch 1 performance: 55.7500 2020-04-28 11:34:38,644 Best performance achived, saving it! 2020-04-28 11:35:25,079 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 4.6917 | loss_kp_2d: 2.14 | loss_kp_3d: 1.53 | loss_shape: 0.01 | loss_pose: 0.37 | e_m_disc_loss: 0.29 | d_m_disc_real: 0.12 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.17 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:35:37,789 (67/67) | batch: 119.7ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 11:36:10,899 Epoch 1, MPJPE: 91.8254, PA-MPJPE: 56.7944, ACCEL: 29.4835, PVE: 108.5572, ACCEL_ERR: 30.2531, 2020-04-28 11:36:11,025 Epoch 2 performance: 56.7944 2020-04-28 11:36:58,171 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 4.8381 | loss_kp_2d: 1.67 | loss_kp_3d: 0.72 | loss_shape: 0.01 | loss_pose: 0.51 | e_m_disc_loss: 0.24 | d_m_disc_real: 0.12 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.20 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:37:10,710 (67/67) | batch: 120.6ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 11:37:43,569 Epoch 2, MPJPE: 92.9256, PA-MPJPE: 57.8990, ACCEL: 31.0366, PVE: 109.2432, ACCEL_ERR: 31.7500, 2020-04-28 11:37:43,745 Epoch 3 performance: 57.8990 2020-04-28 11:38:30,652 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 7.2522 | loss_kp_2d: 1.82 | loss_kp_3d: 0.71 | loss_shape: 0.01 | loss_pose: 0.40 | e_m_disc_loss: 0.28 | d_m_disc_real: 0.09 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.14 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:38:43,548 (67/67) | batch: 123.5ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 11:39:16,966 Epoch 3, MPJPE: 90.3195, PA-MPJPE: 55.3699, ACCEL: 30.3872, PVE: 107.0878, ACCEL_ERR: 31.1134, 2020-04-28 11:39:17,139 Epoch 4 performance: 55.3699 2020-04-28 11:39:17,663 Best performance achived, saving it! 2020-04-28 11:40:04,547 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 7.9543 | loss_kp_2d: 14.69 | loss_kp_3d: 1.25 | loss_shape: 0.01 | loss_pose: 1.29 | e_m_disc_loss: 0.18 | d_m_disc_real: 0.11 | d_m_disc_fake: 0.12 | d_m_disc_loss: 0.23 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:40:17,255 (67/67) | batch: 120.1ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 11:40:50,206 Epoch 4, MPJPE: 91.0215, PA-MPJPE: 56.7844, ACCEL: 28.5512, PVE: 108.0067, ACCEL_ERR: 29.3483, 2020-04-28 11:40:50,333 Epoch 5 performance: 56.7844 2020-04-28 11:41:37,458 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 6.3497 | loss_kp_2d: 1.71 | loss_kp_3d: 0.98 | e_m_disc_loss: 0.22 | d_m_disc_real: 0.12 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.20 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:41:50,052 (67/67) | batch: 119.5ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 11:42:23,029 Epoch 5, MPJPE: 97.6334, PA-MPJPE: 61.5261, ACCEL: 30.8223, PVE: 114.4799, ACCEL_ERR: 31.5960, 2020-04-28 11:42:23,155 Epoch 6 performance: 61.5261 2020-04-28 11:43:09,950 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 8.9018 | loss_kp_2d: 2.01 | loss_kp_3d: 1.40 | loss_shape: 0.01 | loss_pose: 0.45 | e_m_disc_loss: 0.37 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.03 | d_m_disc_loss: 0.09 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:43:22,632 (67/67) | batch: 119.9ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 11:43:55,715 Epoch 6, MPJPE: 93.8067, PA-MPJPE: 57.6596, ACCEL: 27.8746, PVE: 111.8382, ACCEL_ERR: 28.7142, 2020-04-28 11:43:55,842 Epoch 7 performance: 57.6596 2020-04-28 11:44:43,071 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 10.7372 | loss_kp_2d: 1.71 | loss_kp_3d: 2.75 | loss_shape: 0.01 | loss_pose: 0.86 | e_m_disc_loss: 0.23 | d_m_disc_real: 0.10 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.17 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:44:55,870 (67/67) | batch: 120.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 11:45:29,143 Epoch 7, MPJPE: 90.0542, PA-MPJPE: 56.9374, ACCEL: 28.2076, PVE: 107.2786, ACCEL_ERR: 29.0228, 2020-04-28 11:45:29,286 Epoch 8 performance: 56.9374 2020-04-28 11:46:15,986 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 9.4700 | loss_kp_2d: 3.32 | loss_kp_3d: 0.85 | loss_shape: 0.02 | loss_pose: 0.90 | e_m_disc_loss: 0.26 | d_m_disc_real: 0.09 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:46:28,854 (67/67) | batch: 122.0ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 11:47:01,843 Epoch 8, MPJPE: 94.2797, PA-MPJPE: 59.7121, ACCEL: 28.9115, PVE: 111.9045, ACCEL_ERR: 29.7178, 2020-04-28 11:47:01,976 Epoch 9 performance: 59.7121 2020-04-28 11:47:50,123 (500/500) | Total: 0:00:47 | ETA: 0:00:01 | loss: 8.8398 | loss_kp_2d: 1.53 | loss_kp_3d: 0.75 | loss_shape: 0.01 | loss_pose: 0.76 | e_m_disc_loss: 0.23 | d_m_disc_real: 0.09 | d_m_disc_fake: 0.10 | d_m_disc_loss: 0.19 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:48:02,950 (67/67) | batch: 120.8ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 11:48:36,022 Epoch 9, MPJPE: 94.3518, PA-MPJPE: 60.3949, ACCEL: 27.7630, PVE: 111.0637, ACCEL_ERR: 28.6098, 2020-04-28 11:48:36,164 Epoch 10 performance: 60.3949 2020-04-28 11:49:28,555 (500/500) | Total: 0:00:51 | ETA: 0:00:01 | loss: 8.4842 | loss_kp_2d: 3.26 | loss_kp_3d: 1.08 | e_m_disc_loss: 0.21 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.10 | d_m_disc_loss: 0.18 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:49:41,072 (67/67) | batch: 119.9ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 11:50:13,753 Epoch 10, MPJPE: 93.9450, PA-MPJPE: 59.7301, ACCEL: 28.2322, PVE: 111.0503, ACCEL_ERR: 29.0869, 2020-04-28 11:50:13,915 Epoch 11 performance: 59.7301 2020-04-28 11:51:00,829 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 12.5687 | loss_kp_2d: 1.57 | loss_kp_3d: 0.60 | loss_shape: 0.00 | loss_pose: 0.27 | e_m_disc_loss: 0.23 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.09 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:51:13,483 (67/67) | batch: 120.3ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 11:51:46,296 Epoch 11, MPJPE: 96.7613, PA-MPJPE: 60.9168, ACCEL: 28.1490, PVE: 114.0093, ACCEL_ERR: 29.0312, 2020-04-28 11:51:46,461 Epoch 12 performance: 60.9168 2020-04-28 11:52:33,656 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 10.8802 | loss_kp_2d: 1.87 | loss_kp_3d: 1.15 | e_m_disc_loss: 0.25 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.14 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.10 2020-04-28 11:52:46,549 (67/67) | batch: 122.2ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 11:53:19,762 Epoch 12, MPJPE: 92.0588, PA-MPJPE: 58.0508, ACCEL: 27.7194, PVE: 108.7320, ACCEL_ERR: 28.5677, 2020-04-28 11:53:19,960 Epoch 13 performance: 58.0508 2020-04-28 11:54:06,722 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 9.3150 | loss_kp_2d: 1.35 | loss_kp_3d: 1.15 | loss_shape: 0.04 | loss_pose: 1.43 | e_m_disc_loss: 0.37 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.03 | d_m_disc_loss: 0.09 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:54:19,335 (67/67) | batch: 119.7ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 11:54:52,220 Epoch 13, MPJPE: 94.4047, PA-MPJPE: 59.6885, ACCEL: 28.3338, PVE: 111.9611, ACCEL_ERR: 29.1977, 2020-04-28 11:54:52,347 Epoch 14 performance: 59.6885 2020-04-28 11:55:39,235 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 9.3332 | loss_kp_2d: 1.42 | loss_kp_3d: 0.70 | loss_shape: 0.01 | loss_pose: 0.59 | e_m_disc_loss: 0.37 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.09 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:55:51,730 (67/67) | batch: 120.2ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 11:56:24,588 Epoch 14, MPJPE: 94.1678, PA-MPJPE: 59.4965, ACCEL: 28.1984, PVE: 111.6134, ACCEL_ERR: 29.0666, 2020-04-28 11:56:24,750 Epoch 15 performance: 59.4965 2020-04-28 11:57:11,816 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 8.6479 | loss_kp_2d: 1.10 | loss_kp_3d: 0.88 | loss_shape: 0.01 | loss_pose: 0.88 | e_m_disc_loss: 0.35 | d_m_disc_real: 0.03 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.08 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:57:24,681 (67/67) | batch: 121.8ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 11:57:57,377 Epoch 15, MPJPE: 95.3181, PA-MPJPE: 59.7299, ACCEL: 28.2595, PVE: 112.8291, ACCEL_ERR: 29.1237, 2020-04-28 11:57:57,504 Epoch 16 performance: 59.7299 2020-04-28 11:58:44,351 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 13.8388 | loss_kp_2d: 1.77 | loss_kp_3d: 0.89 | loss_shape: 0.07 | loss_pose: 1.16 | e_m_disc_loss: 0.42 | d_m_disc_real: 0.03 | d_m_disc_fake: 0.02 | d_m_disc_loss: 0.05 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 11:58:57,485 (67/67) | batch: 123.8ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 11:59:29,828 Epoch 16, MPJPE: 94.6730, PA-MPJPE: 59.3914, ACCEL: 27.9640, PVE: 111.9995, ACCEL_ERR: 28.8392, 2020-04-28 11:59:30,023 Epoch 17 performance: 59.3914 2020-04-28 12:00:17,015 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 11.0892 | loss_kp_2d: 1.21 | loss_kp_3d: 0.98 | loss_shape: 0.04 | loss_pose: 0.98 | e_m_disc_loss: 0.40 | d_m_disc_real: 0.02 | d_m_disc_fake: 0.02 | d_m_disc_loss: 0.04 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:00:29,634 (67/67) | batch: 119.7ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 12:01:02,604 Epoch 17, MPJPE: 94.8459, PA-MPJPE: 59.3900, ACCEL: 27.8448, PVE: 112.2248, ACCEL_ERR: 28.7284, 2020-04-28 12:01:02,730 Epoch 18 performance: 59.3900 2020-04-28 12:01:54,007 (500/500) | Total: 0:00:50 | ETA: 0:00:01 | loss: 8.9831 | loss_kp_2d: 1.39 | loss_kp_3d: 1.24 | loss_shape: 0.01 | loss_pose: 1.45 | e_m_disc_loss: 0.38 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:02:06,636 (67/67) | batch: 119.3ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 12:02:41,221 Epoch 18, MPJPE: 95.1094, PA-MPJPE: 59.9534, ACCEL: 28.3575, PVE: 112.6448, ACCEL_ERR: 29.2228, 2020-04-28 12:02:41,349 Epoch 19 performance: 59.9534 2020-04-28 12:03:28,241 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 3.7548 | loss_kp_2d: 1.96 | loss_kp_3d: 1.15 | loss_shape: 0.01 | loss_pose: 1.00 | e_m_disc_loss: 0.39 | d_m_disc_real: 0.03 | d_m_disc_fake: 0.04 | d_m_disc_loss: 0.07 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:03:40,978 (67/67) | batch: 120.2ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 12:04:15,542 Epoch 19, MPJPE: 94.5889, PA-MPJPE: 59.8894, ACCEL: 28.4550, PVE: 111.9754, ACCEL_ERR: 29.3059, 2020-04-28 12:04:15,669 Epoch 20 performance: 59.8894 2020-04-28 12:05:07,791 (500/500) | Total: 0:00:51 | ETA: 0:00:01 | loss: 4.2396 | loss_kp_2d: 31.30 | loss_kp_3d: 0.69 | loss_shape: 0.01 | loss_pose: 0.97 | e_m_disc_loss: 0.36 | d_m_disc_real: 0.02 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.07 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:05:20,463 (67/67) | batch: 121.6ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 12:05:53,441 Epoch 20, MPJPE: 94.2937, PA-MPJPE: 59.8265, ACCEL: 28.5905, PVE: 111.7302, ACCEL_ERR: 29.4298, 2020-04-28 12:05:53,617 Epoch 21 performance: 59.8265 2020-04-28 12:06:40,629 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 5.2466 | loss_kp_2d: 1.13 | loss_kp_3d: 1.05 | e_m_disc_loss: 0.41 | d_m_disc_real: 0.02 | d_m_disc_fake: 0.02 | d_m_disc_loss: 0.05 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:06:53,518 (67/67) | batch: 122.6ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 12:07:26,843 Epoch 21, MPJPE: 93.6995, PA-MPJPE: 59.4287, ACCEL: 28.4712, PVE: 111.0612, ACCEL_ERR: 29.3069, 2020-04-28 12:07:26,988 Epoch 22 performance: 59.4287 2020-04-28 12:08:13,843 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 4.6033 | loss_kp_2d: 1.87 | loss_kp_3d: 0.69 | e_m_disc_loss: 0.41 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.02 | d_m_disc_loss: 0.06 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:08:26,825 (67/67) | batch: 124.0ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 12:09:01,331 Epoch 22, MPJPE: 93.7079, PA-MPJPE: 59.4438, ACCEL: 28.4748, PVE: 111.0778, ACCEL_ERR: 29.3107, 2020-04-28 12:09:01,533 Epoch 23 performance: 59.4438 2020-04-28 12:09:48,415 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 6.2962 | loss_kp_2d: 1.23 | loss_kp_3d: 1.17 | loss_shape: 0.01 | loss_pose: 0.34 | e_m_disc_loss: 0.40 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.03 | d_m_disc_loss: 0.07 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:10:01,769 (67/67) | batch: 126.1ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 12:10:38,404 Epoch 23, MPJPE: 93.6566, PA-MPJPE: 59.4125, ACCEL: 28.4584, PVE: 111.0038, ACCEL_ERR: 29.2950, 2020-04-28 12:10:38,559 Epoch 24 performance: 59.4125 2020-04-28 12:11:25,727 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 6.7867 | loss_kp_2d: 3.24 | loss_kp_3d: 0.85 | loss_shape: 0.01 | loss_pose: 0.31 | e_m_disc_loss: 0.37 | d_m_disc_real: 0.02 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.06 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:11:39,161 (67/67) | batch: 128.0ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 12:12:16,121 Epoch 24, MPJPE: 93.8096, PA-MPJPE: 59.5208, ACCEL: 28.4832, PVE: 111.1795, ACCEL_ERR: 29.3203, 2020-04-28 12:12:16,280 Epoch 25 performance: 59.5208 2020-04-28 12:13:03,161 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 7.8271 | loss_kp_2d: 7.42 | loss_kp_3d: 0.58 | e_m_disc_loss: 0.39 | d_m_disc_real: 0.01 | d_m_disc_fake: 0.04 | d_m_disc_loss: 0.05 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:13:15,770 (67/67) | batch: 121.0ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 12:13:48,640 Epoch 25, MPJPE: 93.9059, PA-MPJPE: 59.5756, ACCEL: 28.4973, PVE: 111.3118, ACCEL_ERR: 29.3351, 2020-04-28 12:13:48,815 Epoch 26 performance: 59.5756 2020-04-28 12:14:35,714 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 6.9054 | loss_kp_2d: 1.99 | loss_kp_3d: 0.62 | loss_shape: 0.01 | loss_pose: 1.92 | e_m_disc_loss: 0.45 | d_m_disc_real: 0.03 | d_m_disc_fake: 0.01 | d_m_disc_loss: 0.04 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:14:48,646 (67/67) | batch: 122.8ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 12:15:25,812 Epoch 26, MPJPE: 93.8673, PA-MPJPE: 59.5344, ACCEL: 28.4714, PVE: 111.2644, ACCEL_ERR: 29.3101, 2020-04-28 12:15:25,940 Epoch 27 performance: 59.5344 2020-04-28 12:16:12,825 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 11.6100 | loss_kp_2d: 1.01 | loss_kp_3d: 0.85 | loss_shape: 0.01 | loss_pose: 0.56 | e_m_disc_loss: 0.45 | d_m_disc_real: 0.03 | d_m_disc_fake: 0.02 | d_m_disc_loss: 0.05 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:16:26,284 (67/67) | batch: 127.9ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 12:17:02,938 Epoch 27, MPJPE: 93.8285, PA-MPJPE: 59.4912, ACCEL: 28.4309, PVE: 111.2094, ACCEL_ERR: 29.2708, 2020-04-28 12:17:03,108 Epoch 28 performance: 59.4912 2020-04-28 12:17:49,944 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 10.4088 | loss_kp_2d: 1.39 | loss_kp_3d: 0.57 | loss_shape: 0.01 | loss_pose: 1.38 | e_m_disc_loss: 0.39 | d_m_disc_real: 0.02 | d_m_disc_fake: 0.04 | d_m_disc_loss: 0.06 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:18:03,797 (67/67) | batch: 131.6ms | Total: 0:00:13 | ETA: 0:00:01 2020-04-28 12:18:40,323 Epoch 28, MPJPE: 93.8299, PA-MPJPE: 59.4913, ACCEL: 28.4293, PVE: 111.2096, ACCEL_ERR: 29.2694, 2020-04-28 12:18:40,513 Epoch 29 performance: 59.4913 2020-04-28 12:19:27,187 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 14.7345 | loss_kp_2d: 1.00 | loss_kp_3d: 0.71 | e_m_disc_loss: 0.39 | d_m_disc_real: 0.02 | d_m_disc_fake: 0.03 | d_m_disc_loss: 0.06 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 12:19:41,070 (67/67) | batch: 131.2ms | Total: 0:00:13 | ETA: 0:00:01 2020-04-28 12:20:17,432 Epoch 29, MPJPE: 93.8361, PA-MPJPE: 59.4942, ACCEL: 28.4269, PVE: 111.2153, ACCEL_ERR: 29.2671, 2020-04-28 12:20:17,625 Epoch 30 performance: 59.4942

Could you try disabling training with PoseTrack and PennAction datasets?

And by the way, from the log best performance seems like 55.3

Thank you! May be this log is not the corresponding one. I will try disabling training with PoseTrack and PennAction datasets.

@Frank-Dz only 18 version.

thx！

On 04/28/2020 17:47, Muhammed Kocabas wrote:

@Frank-Dz only 18 version.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

I'm confused about AMASS dataset... On AMASS there are different databases, from below which database we need? Furthermore we need only body data?

Annotation 2020-04-28 180923

I just downloaded all datasets

On 04/28/2020 18:11, Programming lover wrote:

I'm confused about AMASS dataset... On AMASS there are different databases, from below which database we need? Furthermore we need only body data?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

@Frank-Dz Thanks for your prompt response. Regarding Insta_Variety when I download it many videos are missing, may be it is due to that someone deleted videos from Insta. So we need to use pre-processed insta_variety file or we need to update data file using akanazawa method

2020-04-28 17:33:07,802 GPU name -> Tesla V100-SXM2-32GB 2020-04-28 17:33:07,802 GPU feat -> _CudaDeviceProperties(name='Tesla V100-SXM2-32GB', major=7, minor=0, total_memory=32480MB, multi_processor_count=80) 2020-04-28 17:33:07,803 {'CUDNN': CfgNode({'BENCHMARK': True, 'DETERMINISTIC': False, 'ENABLED': True}), 'DATASET': CfgNode({'SEQLEN': 16, 'OVERLAP': 0.5}), 'DEBUG': False, 'DEBUG_FREQ': 5, 'DEVICE': 'cuda', 'EXP_NAME': 'vibe_raw', 'LOGDIR': 'results/vibe_raw/28-04-2020_17-33-07_vibe_raw', 'LOSS': {'D_MOTION_LOSS_W': 0.5, 'KP_2D_W': 300.0, 'KP_3D_W': 300.0, 'POSE_W': 60.0, 'SHAPE_W': 0.06}, 'MODEL': {'TEMPORAL_TYPE': 'gru', 'TGRU': {'ADD_LINEAR': True, 'BIDIRECTIONAL': False, 'HIDDEN_SIZE': 1024, 'NUM_LAYERS': 2, 'RESIDUAL': True}}, 'NUM_WORKERS': 8, 'OUTPUT_DIR': 'results/vibe_raw', 'SEED_VALUE': -1, 'TRAIN': {'BATCH_SIZE': 32, 'DATASETS_2D': ['Insta'], 'DATASETS_3D': ['ThreeDPW', 'MPII3D'], 'DATASET_EVAL': 'ThreeDPW', 'DATA_2D_RATIO': 0.6, 'END_EPOCH': 200, 'GEN_LR': 5e-05, 'GEN_MOMENTUM': 0.9, 'GEN_OPTIM': 'Adam', 'GEN_WD': 0.0, 'LR_PATIENCE': 5, 'MOT_DISCR': {'ATT': {'DROPOUT': 0.2, 'LAYERS': 3, 'SIZE': 1024}, 'FEATURE_POOL': 'attention', 'HIDDEN_SIZE': 1024, 'LR': 0.0001, 'MOMENTUM': 0.9, 'NUM_LAYERS': 2, 'OPTIM': 'Adam', 'UPDATE_STEPS': 1, 'WD': 0.0001}, 'NUM_ITERS_PER_EPOCH': 500, 'PRETRAINED': '', 'PRETRAINED_REGRESSOR': 'data/vibe_data/spin_model_checkpoint.pth.tar', 'RESUME': '', 'START_EPOCH': 0}} 2020-04-28 17:33:51,184 => no checkpoint found at '' 2020-04-28 17:34:37,429 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 3.3052 | loss_kp_2d: 1.50 | loss_kp_3d: 0.92 | e_m_disc_loss: 0.23 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.15 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:34:49,432 (67/67) | batch: 114.1ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:35:22,075 Epoch 0, MPJPE: 88.9655, PA-MPJPE: 54.0784, ACCEL: 27.9975, PVE: 105.8096, ACCEL_ERR: 28.8034, 2020-04-28 17:35:22,201 Epoch 1 performance: 54.0784 2020-04-28 17:35:22,544 Best performance achived, saving it! 2020-04-28 17:36:08,942 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 3.0451 | loss_kp_2d: 1.06 | loss_kp_3d: 0.75 | loss_shape: 0.01 | loss_pose: 0.36 | e_m_disc_loss: 0.39 | d_m_disc_real: 0.18 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.22 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:36:21,418 (67/67) | batch: 118.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:36:54,089 Epoch 1, MPJPE: 90.2474, PA-MPJPE: 55.9259, ACCEL: 28.8877, PVE: 106.9013, ACCEL_ERR: 29.6679, 2020-04-28 17:36:54,216 Epoch 2 performance: 55.9259 2020-04-28 17:37:41,482 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 3.1461 | loss_kp_2d: 0.85 | loss_kp_3d: 0.98 | e_m_disc_loss: 0.17 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.11 | d_m_disc_loss: 0.20 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:37:53,896 (67/67) | batch: 118.2ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:38:26,899 Epoch 2, MPJPE: 88.9309, PA-MPJPE: 55.0614, ACCEL: 27.4448, PVE: 104.3687, ACCEL_ERR: 28.2871, 2020-04-28 17:38:27,024 Epoch 3 performance: 55.0614 2020-04-28 17:39:15,218 (500/500) | Total: 0:00:47 | ETA: 0:00:01 | loss: 3.0387 | loss_kp_2d: 0.70 | loss_kp_3d: 0.71 | e_m_disc_loss: 0.13 | d_m_disc_real: 0.10 | d_m_disc_fake: 0.14 | d_m_disc_loss: 0.25 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:39:27,946 (67/67) | batch: 121.5ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 17:40:00,834 Epoch 3, MPJPE: 87.4501, PA-MPJPE: 54.0522, ACCEL: 27.4908, PVE: 104.7147, ACCEL_ERR: 28.3180, 2020-04-28 17:40:00,960 Epoch 4 performance: 54.0522 2020-04-28 17:40:01,508 Best performance achived, saving it! 2020-04-28 17:40:48,724 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.9539 | loss_kp_2d: 0.66 | loss_kp_3d: 0.69 | loss_shape: 0.03 | loss_pose: 1.93 | e_m_disc_loss: 0.23 | d_m_disc_real: 0.14 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.22 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:41:01,229 (67/67) | batch: 118.7ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:41:34,324 Epoch 4, MPJPE: 88.2255, PA-MPJPE: 54.6182, ACCEL: 27.5513, PVE: 104.2368, ACCEL_ERR: 28.4085, 2020-04-28 17:41:34,466 Epoch 5 performance: 54.6182 2020-04-28 17:42:21,585 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.9196 | loss_kp_2d: 0.66 | loss_kp_3d: 1.19 | loss_shape: 0.01 | loss_pose: 0.36 | e_m_disc_loss: 0.25 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.13 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:42:34,036 (67/67) | batch: 118.8ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:43:06,640 Epoch 5, MPJPE: 88.8532, PA-MPJPE: 56.2239, ACCEL: 27.3844, PVE: 105.3923, ACCEL_ERR: 28.2387, 2020-04-28 17:43:06,768 Epoch 6 performance: 56.2239 2020-04-28 17:43:53,553 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.9362 | loss_kp_2d: 1.73 | loss_kp_3d: 0.83 | loss_shape: 0.01 | loss_pose: 1.63 | e_m_disc_loss: 0.30 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.13 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:44:06,067 (67/67) | batch: 118.3ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:44:38,510 Epoch 6, MPJPE: 88.3176, PA-MPJPE: 55.8464, ACCEL: 26.4748, PVE: 105.4803, ACCEL_ERR: 27.3649, 2020-04-28 17:44:38,713 Epoch 7 performance: 55.8464 2020-04-28 17:45:25,221 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.9353 | loss_kp_2d: 1.00 | loss_kp_3d: 1.12 | loss_shape: 0.02 | loss_pose: 0.75 | e_m_disc_loss: 0.21 | d_m_disc_real: 0.14 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.22 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:45:37,637 (67/67) | batch: 117.2ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:46:10,168 Epoch 7, MPJPE: 88.1955, PA-MPJPE: 55.2046, ACCEL: 26.8387, PVE: 105.0781, ACCEL_ERR: 27.6987, 2020-04-28 17:46:10,294 Epoch 8 performance: 55.2046 2020-04-28 17:46:56,930 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.9401 | loss_kp_2d: 0.94 | loss_kp_3d: 1.26 | e_m_disc_loss: 0.20 | d_m_disc_real: 0.16 | d_m_disc_fake: 0.09 | d_m_disc_loss: 0.25 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:47:09,368 (67/67) | batch: 118.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:47:42,894 Epoch 8, MPJPE: 89.4215, PA-MPJPE: 55.9550, ACCEL: 26.3479, PVE: 106.6388, ACCEL_ERR: 27.2342, 2020-04-28 17:47:43,039 Epoch 9 performance: 55.9550 2020-04-28 17:48:29,652 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.8680 | loss_kp_2d: 0.68 | loss_kp_3d: 1.04 | e_m_disc_loss: 0.20 | d_m_disc_real: 0.16 | d_m_disc_fake: 0.09 | d_m_disc_loss: 0.25 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:48:42,464 (67/67) | batch: 121.5ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 17:49:15,286 Epoch 9, MPJPE: 88.1164, PA-MPJPE: 54.8346, ACCEL: 26.2051, PVE: 103.8222, ACCEL_ERR: 27.1294, 2020-04-28 17:49:15,415 Epoch 10 performance: 54.8346 2020-04-28 17:50:06,119 (500/500) | Total: 0:00:50 | ETA: 0:00:01 | loss: 2.7482 | loss_kp_2d: 0.49 | loss_kp_3d: 0.75 | loss_shape: 0.00 | loss_pose: 0.81 | e_m_disc_loss: 0.23 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.15 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:50:18,572 (67/67) | batch: 118.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:50:51,661 Epoch 10, MPJPE: 86.4797, PA-MPJPE: 55.2742, ACCEL: 25.8963, PVE: 102.9881, ACCEL_ERR: 26.8101, 2020-04-28 17:50:51,805 Epoch 11 performance: 55.2742 2020-04-28 17:51:38,600 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6959 | loss_kp_2d: 0.85 | loss_kp_3d: 1.00 | loss_shape: 0.01 | loss_pose: 0.32 | e_m_disc_loss: 0.24 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.14 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:51:51,210 (67/67) | batch: 120.1ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 17:52:25,249 Epoch 11, MPJPE: 85.8728, PA-MPJPE: 54.6931, ACCEL: 25.5772, PVE: 102.0871, ACCEL_ERR: 26.5127, 2020-04-28 17:52:25,406 Epoch 12 performance: 54.6931 2020-04-28 17:53:12,202 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6467 | loss_kp_2d: 0.83 | loss_kp_3d: 1.09 | loss_shape: 0.03 | loss_pose: 0.58 | e_m_disc_loss: 0.22 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.10 | d_m_disc_loss: 0.15 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:53:24,947 (67/67) | batch: 120.7ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 17:53:58,189 Epoch 12, MPJPE: 86.3866, PA-MPJPE: 55.1251, ACCEL: 25.6505, PVE: 102.5202, ACCEL_ERR: 26.5789, 2020-04-28 17:53:58,334 Epoch 13 performance: 55.1251 2020-04-28 17:54:50,234 (500/500) | Total: 0:00:51 | ETA: 0:00:01 | loss: 2.6951 | loss_kp_2d: 1.10 | loss_kp_3d: 0.71 | loss_shape: 0.01 | loss_pose: 0.37 | e_m_disc_loss: 0.25 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:55:02,641 (67/67) | batch: 118.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:55:35,460 Epoch 13, MPJPE: 86.4423, PA-MPJPE: 55.2600, ACCEL: 25.9393, PVE: 102.5090, ACCEL_ERR: 26.8482, 2020-04-28 17:55:35,645 Epoch 14 performance: 55.2600 2020-04-28 17:56:22,458 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6767 | loss_kp_2d: 1.04 | loss_kp_3d: 0.78 | e_m_disc_loss: 0.24 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.09 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:56:34,586 (67/67) | batch: 116.9ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:57:07,620 Epoch 14, MPJPE: 86.3996, PA-MPJPE: 54.9225, ACCEL: 25.5423, PVE: 102.4451, ACCEL_ERR: 26.4808, 2020-04-28 17:57:07,770 Epoch 15 performance: 54.9225 2020-04-28 17:57:54,521 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7793 | loss_kp_2d: 0.73 | loss_kp_3d: 1.50 | e_m_disc_loss: 0.27 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.12 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:58:06,833 (67/67) | batch: 118.2ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:58:39,767 Epoch 15, MPJPE: 86.4569, PA-MPJPE: 54.6776, ACCEL: 25.5324, PVE: 102.3305, ACCEL_ERR: 26.4675, 2020-04-28 17:58:39,950 Epoch 16 performance: 54.6776 2020-04-28 17:59:26,793 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7049 | loss_kp_2d: 0.82 | loss_kp_3d: 0.90 | loss_shape: 0.01 | loss_pose: 0.68 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.03 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:59:39,277 (67/67) | batch: 119.4ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:00:11,704 Epoch 16, MPJPE: 86.5707, PA-MPJPE: 54.8733, ACCEL: 25.4781, PVE: 102.3858, ACCEL_ERR: 26.4126, 2020-04-28 18:00:11,883 Epoch 17 performance: 54.8733 2020-04-28 18:00:58,426 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6740 | loss_kp_2d: 1.11 | loss_kp_3d: 0.86 | e_m_disc_loss: 0.36 | d_m_disc_real: 0.09 | d_m_disc_fake: 0.04 | d_m_disc_loss: 0.14 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:01:10,849 (67/67) | batch: 118.1ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:01:43,817 Epoch 17, MPJPE: 86.5733, PA-MPJPE: 54.7636, ACCEL: 25.4780, PVE: 102.4244, ACCEL_ERR: 26.4129, 2020-04-28 18:01:43,945 Epoch 18 performance: 54.7636 2020-04-28 18:02:30,809 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6949 | loss_kp_2d: 0.70 | loss_kp_3d: 0.76 | loss_shape: 0.00 | loss_pose: 0.81 | e_m_disc_loss: 0.36 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:02:43,000 (67/67) | batch: 117.5ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:03:15,695 Epoch 18, MPJPE: 86.5603, PA-MPJPE: 54.7283, ACCEL: 25.4308, PVE: 102.4448, ACCEL_ERR: 26.3688, 2020-04-28 18:03:15,845 Epoch 19 performance: 54.7283 2020-04-28 18:04:03,283 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6551 | loss_kp_2d: 0.65 | loss_kp_3d: 0.74 | e_m_disc_loss: 0.29 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:04:15,465 (67/67) | batch: 117.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:04:47,963 Epoch 19, MPJPE: 86.5021, PA-MPJPE: 54.6963, ACCEL: 25.4630, PVE: 102.3905, ACCEL_ERR: 26.3996, 2020-04-28 18:04:48,115 Epoch 20 performance: 54.6963 2020-04-28 18:05:39,558 (500/500) | Total: 0:00:50 | ETA: 0:00:01 | loss: 2.7341 | loss_kp_2d: 1.15 | loss_kp_3d: 0.67 | e_m_disc_loss: 0.29 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.09 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:05:52,395 (67/67) | batch: 122.7ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:06:25,096 Epoch 20, MPJPE: 86.5212, PA-MPJPE: 54.6335, ACCEL: 25.4595, PVE: 102.3775, ACCEL_ERR: 26.3960, 2020-04-28 18:06:25,223 Epoch 21 performance: 54.6335 2020-04-28 18:07:12,379 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7275 | loss_kp_2d: 1.02 | loss_kp_3d: 0.84 | loss_shape: 0.06 | loss_pose: 1.16 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:07:24,547 (67/67) | batch: 117.3ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:07:57,038 Epoch 21, MPJPE: 86.5159, PA-MPJPE: 54.7687, ACCEL: 25.5565, PVE: 102.3806, ACCEL_ERR: 26.4873, 2020-04-28 18:07:57,190 Epoch 22 performance: 54.7687 2020-04-28 18:08:44,248 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6775 | loss_kp_2d: 0.72 | loss_kp_3d: 0.76 | loss_shape: 0.04 | loss_pose: 1.37 | e_m_disc_loss: 0.32 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.15 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:08:56,404 (67/67) | batch: 116.7ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:09:29,021 Epoch 22, MPJPE: 86.4920, PA-MPJPE: 54.7528, ACCEL: 25.5490, PVE: 102.3578, ACCEL_ERR: 26.4807, 2020-04-28 18:09:29,172 Epoch 23 performance: 54.7528 2020-04-28 18:10:15,994 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6667 | loss_kp_2d: 0.70 | loss_kp_3d: 0.88 | e_m_disc_loss: 0.32 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:10:28,157 (67/67) | batch: 116.8ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:11:00,603 Epoch 23, MPJPE: 86.4850, PA-MPJPE: 54.7576, ACCEL: 25.5387, PVE: 102.3518, ACCEL_ERR: 26.4709, 2020-04-28 18:11:00,783 Epoch 24 performance: 54.7576 2020-04-28 18:11:47,945 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6311 | loss_kp_2d: 0.72 | loss_kp_3d: 0.65 | loss_shape: 0.00 | loss_pose: 0.43 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.10 2020-04-28 18:12:00,262 (67/67) | batch: 117.2ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:12:32,859 Epoch 24, MPJPE: 86.4939, PA-MPJPE: 54.7664, ACCEL: 25.5310, PVE: 102.3616, ACCEL_ERR: 26.4637, 2020-04-28 18:12:33,011 Epoch 25 performance: 54.7664 2020-04-28 18:13:20,511 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7219 | loss_kp_2d: 0.74 | loss_kp_3d: 0.47 | loss_shape: 0.00 | loss_pose: 0.28 | e_m_disc_loss: 0.37 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:13:33,380 (67/67) | batch: 121.6ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:14:05,994 Epoch 25, MPJPE: 86.4563, PA-MPJPE: 54.7460, ACCEL: 25.5327, PVE: 102.3399, ACCEL_ERR: 26.4654, 2020-04-28 18:14:06,122 Epoch 26 performance: 54.7460 2020-04-28 18:14:53,012 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6909 | loss_kp_2d: 0.88 | loss_kp_3d: 0.90 | e_m_disc_loss: 0.29 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.13 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:15:05,170 (67/67) | batch: 117.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:15:37,938 Epoch 26, MPJPE: 86.4457, PA-MPJPE: 54.7430, ACCEL: 25.5239, PVE: 102.3337, ACCEL_ERR: 26.4575, 2020-04-28 18:15:38,133 Epoch 27 performance: 54.7430 2020-04-28 18:16:28,684 (500/500) | Total: 0:00:50 | ETA: 0:00:01 | loss: 2.6425 | loss_kp_2d: 0.74 | loss_kp_3d: 0.87 | loss_shape: 0.00 | loss_pose: 0.79 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:16:41,071 (67/67) | batch: 117.6ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:17:14,417 Epoch 27, MPJPE: 86.4483, PA-MPJPE: 54.7389, ACCEL: 25.5125, PVE: 102.3312, ACCEL_ERR: 26.4467, 2020-04-28 18:17:14,549 Epoch 28 performance: 54.7389 2020-04-28 18:18:01,348 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7038 | loss_kp_2d: 0.55 | loss_kp_3d: 0.84 | e_m_disc_loss: 0.34 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.12 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:18:13,642 (67/67) | batch: 118.5ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:18:46,395 Epoch 28, MPJPE: 86.4503, PA-MPJPE: 54.7388, ACCEL: 25.5113, PVE: 102.3329, ACCEL_ERR: 26.4455, 2020-04-28 18:18:46,561 Epoch 29 performance: 54.7388 2020-04-28 18:19:33,977 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7691 | loss_kp_2d: 0.76 | loss_kp_3d: 0.64 | loss_shape: 0.01 | loss_pose: 0.45 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:19:46,518 (67/67) | batch: 119.6ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:20:19,760 Epoch 29, MPJPE: 86.4516, PA-MPJPE: 54.7397, ACCEL: 25.5113, PVE: 102.3342, ACCEL_ERR: 26.4456, 2020-04-28 18:20:19,889 Epoch 30 performance: 54.7397 2020-04-28 18:21:06,507 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7000 | loss_kp_2d: 0.87 | loss_kp_3d: 0.79 | loss_shape: 0.00 | loss_pose: 0.51 | e_m_disc_loss: 0.31 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:21:18,808 (67/67) | batch: 116.9ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:21:51,760 Epoch 30, MPJPE: 86.4529, PA-MPJPE: 54.7399, ACCEL: 25.5112, PVE: 102.3347, ACCEL_ERR: 26.4454, 2020-04-28 18:21:51,909 Epoch 31 performance: 54.7399 2020-04-28 18:22:43,510 (500/500) | Total: 0:00:51 | ETA: 0:00:01 | loss: 2.7707 | loss_kp_2d: 1.01 | loss_kp_3d: 1.07 | loss_shape: 0.01 | loss_pose: 0.31 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.12 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:22:56,218 (67/67) | batch: 120.1ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:23:29,423 Epoch 31, MPJPE: 86.4545, PA-MPJPE: 54.7412, ACCEL: 25.5098, PVE: 102.3357, ACCEL_ERR: 26.4442, 2020-04-28 18:23:29,553 Epoch 32 performance: 54.7412 2020-04-28 18:24:16,216 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7585 | loss_kp_2d: 0.95 | loss_kp_3d: 0.80 | loss_shape: 0.01 | loss_pose: 0.79 | e_m_disc_loss: 0.27 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.12 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:24:28,556 (67/67) | batch: 117.6ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:25:01,837 Epoch 32, MPJPE: 86.4553, PA-MPJPE: 54.7427, ACCEL: 25.5109, PVE: 102.3368, ACCEL_ERR: 26.4452, 2020-04-28 18:25:01,980 Epoch 33 performance: 54.7427 2020-04-28 18:25:48,806 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6797 | loss_kp_2d: 0.62 | loss_kp_3d: 0.64 | loss_shape: 0.01 | loss_pose: 0.44 | e_m_disc_loss: 0.27 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.14 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:26:01,470 (67/67) | batch: 120.1ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:26:34,403 Epoch 33, MPJPE: 86.4581, PA-MPJPE: 54.7434, ACCEL: 25.5093, PVE: 102.3390, ACCEL_ERR: 26.4437, 2020-04-28 18:26:34,562 Epoch 34 performance: 54.7434 2020-04-28 18:27:21,259 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7101 | loss_kp_2d: 0.57 | loss_kp_3d: 0.60 | loss_shape: 0.01 | loss_pose: 0.76 | e_m_disc_loss: 0.26 | d_m_disc_real: 0.09 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:27:33,998 (67/67) | batch: 120.8ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:28:07,159 Epoch 34, MPJPE: 86.4568, PA-MPJPE: 54.7435, ACCEL: 25.5098, PVE: 102.3383, ACCEL_ERR: 26.4441, 2020-04-28 18:28:07,290 Epoch 35 performance: 54.7435 2020-04-28 18:28:53,883 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6428 | loss_kp_2d: 0.86 | loss_kp_3d: 0.59 | e_m_disc_loss: 0.30 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:29:06,128 (67/67) | batch: 116.3ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:29:39,133 Epoch 35, MPJPE: 86.4554, PA-MPJPE: 54.7427, ACCEL: 25.5101, PVE: 102.3382, ACCEL_ERR: 26.4445, 2020-04-28 18:29:39,281 Epoch 36 performance: 54.7427 2020-04-28 18:30:26,179 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7171 | loss_kp_2d: 1.22 | loss_kp_3d: 1.02 | loss_shape: 0.02 | loss_pose: 0.58 | e_m_disc_loss: 0.40 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.03 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:30:38,404 (67/67) | batch: 117.2ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:31:11,132 Epoch 36, MPJPE: 86.4577, PA-MPJPE: 54.7441, ACCEL: 25.5088, PVE: 102.3392, ACCEL_ERR: 26.4432, 2020-04-28 18:31:11,318 Epoch 37 performance: 54.7441 2020-04-28 18:31:58,223 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.5947 | loss_kp_2d: 1.28 | loss_kp_3d: 0.95 | loss_shape: 0.02 | loss_pose: 0.94 | e_m_disc_loss: 0.34 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:32:10,617 (67/67) | batch: 117.9ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:32:43,657 Epoch 37, MPJPE: 86.4566, PA-MPJPE: 54.7438, ACCEL: 25.5082, PVE: 102.3384, ACCEL_ERR: 26.4426, 2020-04-28 18:32:43,811 Epoch 38 performance: 54.7438 2020-04-28 18:33:31,157 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6832 | loss_kp_2d: 0.49 | loss_kp_3d: 0.70 | loss_shape: 0.02 | loss_pose: 1.42 | e_m_disc_loss: 0.27 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:33:44,135 (67/67) | batch: 123.5ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:34:17,023 Epoch 38, MPJPE: 86.4575, PA-MPJPE: 54.7450, ACCEL: 25.5070, PVE: 102.3389, ACCEL_ERR: 26.4415, 2020-04-28 18:34:17,169 Epoch 39 performance: 54.7450 2020-04-28 18:35:03,834 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6798 | loss_kp_2d: 0.99 | loss_kp_3d: 0.76 | loss_shape: 0.01 | loss_pose: 0.50 | e_m_disc_loss: 0.30 | d_m_disc_real: 0.09 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:35:16,499 (67/67) | batch: 120.4ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:35:49,399 Epoch 39, MPJPE: 86.4579, PA-MPJPE: 54.7455, ACCEL: 25.5073, PVE: 102.3389, ACCEL_ERR: 26.4418, 2020-04-28 18:35:49,545 Epoch 40 performance: 54.7455 2020-04-28 18:36:37,016 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6764 | loss_kp_2d: 0.83 | loss_kp_3d: 0.64 | loss_shape: 0.02 | loss_pose: 0.74 | e_m_disc_loss: 0.28 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.13 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.10 2020-04-28 18:36:49,772 (67/67) | batch: 120.7ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:37:22,523 Epoch 40, MPJPE: 86.4577, PA-MPJPE: 54.7448, ACCEL: 25.5076, PVE: 102.3405, ACCEL_ERR: 26.4421, 2020-04-28 18:37:22,670 Epoch 41 performance: 54.7448 2020-04-28 18:38:19,826 (500/500) | Total: 0:00:56 | ETA: 0:00:01 | loss: 2.6005 | loss_kp_2d: 1.05 | loss_kp_3d: 0.59 | loss_shape: 0.01 | loss_pose: 0.57 | e_m_disc_loss: 0.36 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:38:32,518 (67/67) | batch: 119.9ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:39:05,219 Epoch 41, MPJPE: 86.4564, PA-MPJPE: 54.7423, ACCEL: 25.5072, PVE: 102.3389, ACCEL_ERR: 26.4417, 2020-04-28 18:39:05,347 Epoch 42 performance: 54.7423 2020-04-28 18:39:52,108 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7966 | loss_kp_2d: 0.80 | loss_kp_3d: 0.59 | loss_shape: 0.01 | loss_pose: 0.56 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:40:04,600 (67/67) | batch: 119.1ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:40:37,272 Epoch 42, MPJPE: 86.4580, PA-MPJPE: 54.7440, ACCEL: 25.5074, PVE: 102.3408, ACCEL_ERR: 26.4419, 2020-04-28 18:40:37,399 Epoch 43 performance: 54.7440 2020-04-28 18:41:24,532 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7288 | loss_kp_2d: 1.10 | loss_kp_3d: 2.34 | e_m_disc_loss: 0.28 | d_m_disc_real: 0.03 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:41:37,029 (67/67) | batch: 118.4ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:42:10,256 Epoch 43, MPJPE: 86.4579, PA-MPJPE: 54.7424, ACCEL: 25.5061, PVE: 102.3399, ACCEL_ERR: 26.4407, 2020-04-28 18:42:10,401 Epoch 44 performance: 54.7424 2020-04-28 18:42:57,285 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6371 | loss_kp_2d: 0.73 | loss_kp_3d: 0.86 | loss_shape: 0.01 | loss_pose: 0.99 | e_m_disc_loss: 0.34 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:43:10,172 (67/67) | batch: 121.7ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:43:43,008 Epoch 44, MPJPE: 86.4590, PA-MPJPE: 54.7436, ACCEL: 25.5054, PVE: 102.3400, ACCEL_ERR: 26.4400, 2020-04-28 18:43:43,136 Epoch 45 performance: 54.7436 2020-04-28 18:44:30,157 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7158 | loss_kp_2d: 0.64 | loss_kp_3d: 0.77 | loss_shape: 0.01 | loss_pose: 1.02 | e_m_disc_loss: 0.31 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.12 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:44:42,585 (67/67) | batch: 120.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:45:15,383 Epoch 45, MPJPE: 86.4590, PA-MPJPE: 54.7435, ACCEL: 25.5045, PVE: 102.3402, ACCEL_ERR: 26.4392, 2020-04-28 18:45:15,546 Epoch 46 performance: 54.7435 2020-04-28 18:46:02,232 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6847 | loss_kp_2d: 0.74 | loss_kp_3d: 0.58 | loss_shape: 0.01 | loss_pose: 0.76 | e_m_disc_loss: 0.31 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:46:14,368 (67/67) | batch: 116.4ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:46:47,041 Epoch 46, MPJPE: 86.4583, PA-MPJPE: 54.7435, ACCEL: 25.5045, PVE: 102.3402, ACCEL_ERR: 26.4391, 2020-04-28 18:46:47,202 Epoch 47 performance: 54.7435 2020-04-28 18:47:34,374 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6367 | loss_kp_2d: 0.92 | loss_kp_3d: 0.58 | loss_shape: 0.00 | loss_pose: 0.33 | e_m_disc_loss: 0.32 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.09 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:47:47,335 (67/67) | batch: 122.2ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:48:20,398 Epoch 47, MPJPE: 86.4597, PA-MPJPE: 54.7439, ACCEL: 25.5030, PVE: 102.3411, ACCEL_ERR: 26.4377, 2020-04-28 18:48:20,595 Epoch 48 performance: 54.7439 2020-04-28 18:49:07,432 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6562 | loss_kp_2d: 0.91 | loss_kp_3d: 0.94 | loss_shape: 0.00 | loss_pose: 1.43 | e_m_disc_loss: 0.31 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.13 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:49:19,865 (67/67) | batch: 118.2ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:49:52,430 Epoch 48, MPJPE: 86.4590, PA-MPJPE: 54.7430, ACCEL: 25.5033, PVE: 102.3402, ACCEL_ERR: 26.4380, 2020-04-28 18:49:52,557 Epoch 49 performance: 54.7430 2020-04-28 18:50:39,425 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6129 | loss_kp_2d: 1.01 | loss_kp_3d: 0.71 | loss_shape: 0.01 | loss_pose: 0.44 | e_m_disc_loss: 0.38 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.13 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:50:51,909 (67/67) | batch: 118.6ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:51:25,130 Epoch 49, MPJPE: 86.4603, PA-MPJPE: 54.7443, ACCEL: 25.5021, PVE: 102.3413, ACCEL_ERR: 26.4369, 2020-04-28 18:51:25,271 Epoch 50 performance: 54.7443

When I disable those 2D datasets, I got eval result like 54.0, which still have 2mm gap with the paper. And I find that the eval error increases, as the training goes. This situation is common?

@Frank-Dz Thanks for your prompt response. Regarding Insta_Variety when I download it many videos are missing, may be it is due to that someone deleted videos from Insta. So we need to use pre-processed insta_variety file or we need to update data file using akanazawa method

Hi~Sorry for bothering. I am sorry, the data I can not smoothly download is insta_variety via akanazawa's scripts. : ( So if it is possible to share the complete insta_variety data you downloaded to me via some online clouds? Thanks!

Forget the following: My network is quite slow while downloading the mpi_inf_3dhp. If it is possible to share the mpi_inf_3dhp data you downloaded to me via some online clouds? Thanks very much!

@Frank-Dz Thanks for your prompt response. Regarding Insta_Variety when I download it many videos are missing, may be it is due to that someone deleted videos from Insta. So we need to use pre-processed insta_variety file or we need to update data file using akanazawa method

@ZephirGe Hi, did you meet the data problem with Insta_Variety during your training process? Thanks!

I'm confused about AMASS dataset... On AMASS there are different databases, from below which database we need? Furthermore we need only body data?

No, AMASS in a dataset that unifies all those data. It has its own website where you can find the data and download them.

@Frank-Dz I'm still downloading insta_variety so I'll try to train after downloading it. Share your email so that I will share mpi_inf_3dhp dataset. Kindly read license file in this zip to use mpi_inf_3dhp dataset.

I'm confused about AMASS dataset... On AMASS there are different databases, from below which database we need? Furthermore we need only body data?

No, AMASS in a dataset that unifies all those data. It has its own website where you can find the data and download them.

@athn-nik I'm following same website that you suggested, In this website I found above databases mentioned in screenshot. Kindly have a look again and help us to find right dataset.

Hi~@lan786 Thank you very much! I have downloaded mpi_inf_3dhp dataset. (spend a very long time...) Currently what I need is insta_variety downloaded via akanazawa's scripts. Can you send it to me if it is possible? Mail: frank33dz@gmail.com Thanks again! : D

@Frank-Dz I tried many times to download insta_variety but it is failed after 2 days 😆 Even I don't know about size of this dataset. Failed on 23GB.

@Frank-Dz I tried many times to download insta_variety but it is failed after 2 days Even I don't know about size of this dataset. Failed on 23GB.

Yep. I used akanazawa's script and started download videos:

python download_insta_variety.py --savedir /path/to/your/save/directory

on https://github.com/akanazawa/human_dynamics/blob/master/doc/insta_variety.md#pre-processed-tfrecords However, the connection is often interrupted, causing my download to fail. : (

@Frank-Dz I tried many times to download insta_variety but it is failed after 2 days Even I don't know about size of this dataset. Failed on 23GB.

Do we really need to preprocess the data by ourselves?

I asked, but got no reply.

@Frank-Dz Thanks for your prompt response. Regarding Insta_Variety when I download it many videos are missing, may be it is due to that someone deleted videos from Insta. So we need to use pre-processed insta_variety file or we need to update data file using akanazawa method

@ZephirGe Hi, did you meet the data problem with Insta_Variety during your training process? Thanks!

@Frank-Dz I will modify downloading code or json file to avoid re-downloading those files which are already downloaded then I will share with you. I don't think so we need to pre-process by our-self, authors of VIBE also shared pre-processed data file. But if we need to add some our own videos then we need to process it that's why I'm trying to download it.

@Frank-Dz I will modify downloading code or json file to avoid re-downloading those files which are already downloaded then I will share with you. I don't think so we need to pre-process by our-self, authors of VIBE also shared pre-processed data file. But if we need to add some our own videos then we need to process it that's why I'm trying to download it.

@lan786

Regarding Insta_Variety when I download it many videos are missing, may be it is due to that someone deleted videos from Insta. So we need to use pre-processed insta_variety file or we need to update data file using akanazawa method

That means if we just wanna train the code, the data author shared is enough "For your convenience, we uploaded the preprocessed InstaVariety data here (size: 18 GB)." ? (Sorry, there are something wrong with my GPU so I haven't tested it. )

@Frank-Dz I will modify downloading code or json file to avoid re-downloading those files which are already downloaded then I will share with you. I don't think so we need to pre-process by our-self, authors of VIBE also shared pre-processed data file. But if we need to add some our own videos then we need to process it that's why I'm trying to download it.

@lan786

Regarding Insta_Variety when I download it many videos are missing, may be it is due to that someone deleted videos from Insta. So we need to use pre-processed insta_variety file or we need to update data file using akanazawa method

That means if we just wanna train the code, the data author shared is enough "For your convenience, we uploaded the preprocessed InstaVariety data here (size: 18 GB)." ? (Sorry, there are something wrong with my GPU so I haven't tested it. )

yes you can use preprocessed InstaVariety but still I'm not sure that we need original dataset along with videos/frames because I didn't read the code.

@Frank-Dz I will modify downloading code or json file to avoid re-downloading those files which are already downloaded then I will share with you. I don't think so we need to pre-process by our-self, authors of VIBE also shared pre-processed data file. But if we need to add some our own videos then we need to process it that's why I'm trying to download it.

@lan786

Regarding Insta_Variety when I download it many videos are missing, may be it is due to that someone deleted videos from Insta. So we need to use pre-processed insta_variety file or we need to update data file using akanazawa method

That means if we just wanna train the code, the data author shared is enough "For your convenience, we uploaded the preprocessed InstaVariety data here (size: 18 GB)." ? (Sorry, there are something wrong with my GPU so I haven't tested it. )

yes you can use preprocessed InstaVariety but still I'm not sure that we need original dataset along with videos/frames because I didn't read the code.

Thanks! I will test it ASAP. Thank you for your kind help again!

@Frank-Dz I will modify downloading code or json file to avoid re-downloading those files which are already downloaded then I will share with you. I don't think so we need to pre-process by our-self, authors of VIBE also shared pre-processed data file. But if we need to add some our own videos then we need to process it that's why I'm trying to download it.

@lan786

Regarding Insta_Variety when I download it many videos are missing, may be it is due to that someone deleted videos from Insta. So we need to use pre-processed insta_variety file or we need to update data file using akanazawa method

That means if we just wanna train the code, the data author shared is enough "For your convenience, we uploaded the preprocessed InstaVariety data here (size: 18 GB)." ? (Sorry, there are something wrong with my GPU so I haven't tested it. )

yes you can use preprocessed InstaVariety but still I'm not sure that we need original dataset along with videos/frames because I didn't read the code.

Thanks! I will test it ASAP. Thank you for your kind help again!

Also let us know about it after testing. Thanks

I'm confused about AMASS dataset... On AMASS there are different databases, from below which database we need? Furthermore we need only body data?

No, AMASS in a dataset that unifies all those data. It has its own website where you can find the data and download them.

@athn-nik I'm following same website that you suggested, In this website I found above databases mentioned in screenshot. Kindly have a look again and help us to find right dataset.

All of them. Amass is the union of all those MoCap data to a single dataset in SMPL/SMPL-H format. Please refer to the original dataset for more details. Yes, you need the body data. Body pose parameters.

@Frank-Dz Thanks for your prompt response. Regarding Insta_Variety when I download it many videos are missing, may be it is due to that someone deleted videos from Insta. So we need to use pre-processed insta_variety file or we need to update data file using akanazawa method

I just use the preprocessed data...

2020-04-28 17:33:07,802 GPU name -> Tesla V100-SXM2-32GB 2020-04-28 17:33:07,802 GPU feat -> _CudaDeviceProperties(name='Tesla V100-SXM2-32GB', major=7, minor=0, total_memory=32480MB, multi_processor_count=80) 2020-04-28 17:33:07,803 {'CUDNN': CfgNode({'BENCHMARK': True, 'DETERMINISTIC': False, 'ENABLED': True}), 'DATASET': CfgNode({'SEQLEN': 16, 'OVERLAP': 0.5}), 'DEBUG': False, 'DEBUG_FREQ': 5, 'DEVICE': 'cuda', 'EXP_NAME': 'vibe_raw', 'LOGDIR': 'results/vibe_raw/28-04-2020_17-33-07_vibe_raw', 'LOSS': {'D_MOTION_LOSS_W': 0.5, 'KP_2D_W': 300.0, 'KP_3D_W': 300.0, 'POSE_W': 60.0, 'SHAPE_W': 0.06}, 'MODEL': {'TEMPORAL_TYPE': 'gru', 'TGRU': {'ADD_LINEAR': True, 'BIDIRECTIONAL': False, 'HIDDEN_SIZE': 1024, 'NUM_LAYERS': 2, 'RESIDUAL': True}}, 'NUM_WORKERS': 8, 'OUTPUT_DIR': 'results/vibe_raw', 'SEED_VALUE': -1, 'TRAIN': {'BATCH_SIZE': 32, 'DATASETS_2D': ['Insta'], 'DATASETS_3D': ['ThreeDPW', 'MPII3D'], 'DATASET_EVAL': 'ThreeDPW', 'DATA_2D_RATIO': 0.6, 'END_EPOCH': 200, 'GEN_LR': 5e-05, 'GEN_MOMENTUM': 0.9, 'GEN_OPTIM': 'Adam', 'GEN_WD': 0.0, 'LR_PATIENCE': 5, 'MOT_DISCR': {'ATT': {'DROPOUT': 0.2, 'LAYERS': 3, 'SIZE': 1024}, 'FEATURE_POOL': 'attention', 'HIDDEN_SIZE': 1024, 'LR': 0.0001, 'MOMENTUM': 0.9, 'NUM_LAYERS': 2, 'OPTIM': 'Adam', 'UPDATE_STEPS': 1, 'WD': 0.0001}, 'NUM_ITERS_PER_EPOCH': 500, 'PRETRAINED': '', 'PRETRAINED_REGRESSOR': 'data/vibe_data/spin_model_checkpoint.pth.tar', 'RESUME': '', 'START_EPOCH': 0}} 2020-04-28 17:33:51,184 => no checkpoint found at '' 2020-04-28 17:34:37,429 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 3.3052 | loss_kp_2d: 1.50 | loss_kp_3d: 0.92 | e_m_disc_loss: 0.23 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.15 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:34:49,432 (67/67) | batch: 114.1ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:35:22,075 Epoch 0, MPJPE: 88.9655, PA-MPJPE: 54.0784, ACCEL: 27.9975, PVE: 105.8096, ACCEL_ERR: 28.8034, 2020-04-28 17:35:22,201 Epoch 1 performance: 54.0784 2020-04-28 17:35:22,544 Best performance achived, saving it! 2020-04-28 17:36:08,942 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 3.0451 | loss_kp_2d: 1.06 | loss_kp_3d: 0.75 | loss_shape: 0.01 | loss_pose: 0.36 | e_m_disc_loss: 0.39 | d_m_disc_real: 0.18 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.22 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:36:21,418 (67/67) | batch: 118.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:36:54,089 Epoch 1, MPJPE: 90.2474, PA-MPJPE: 55.9259, ACCEL: 28.8877, PVE: 106.9013, ACCEL_ERR: 29.6679, 2020-04-28 17:36:54,216 Epoch 2 performance: 55.9259 2020-04-28 17:37:41,482 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 3.1461 | loss_kp_2d: 0.85 | loss_kp_3d: 0.98 | e_m_disc_loss: 0.17 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.11 | d_m_disc_loss: 0.20 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:37:53,896 (67/67) | batch: 118.2ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:38:26,899 Epoch 2, MPJPE: 88.9309, PA-MPJPE: 55.0614, ACCEL: 27.4448, PVE: 104.3687, ACCEL_ERR: 28.2871, 2020-04-28 17:38:27,024 Epoch 3 performance: 55.0614 2020-04-28 17:39:15,218 (500/500) | Total: 0:00:47 | ETA: 0:00:01 | loss: 3.0387 | loss_kp_2d: 0.70 | loss_kp_3d: 0.71 | e_m_disc_loss: 0.13 | d_m_disc_real: 0.10 | d_m_disc_fake: 0.14 | d_m_disc_loss: 0.25 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:39:27,946 (67/67) | batch: 121.5ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 17:40:00,834 Epoch 3, MPJPE: 87.4501, PA-MPJPE: 54.0522, ACCEL: 27.4908, PVE: 104.7147, ACCEL_ERR: 28.3180, 2020-04-28 17:40:00,960 Epoch 4 performance: 54.0522 2020-04-28 17:40:01,508 Best performance achived, saving it! 2020-04-28 17:40:48,724 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.9539 | loss_kp_2d: 0.66 | loss_kp_3d: 0.69 | loss_shape: 0.03 | loss_pose: 1.93 | e_m_disc_loss: 0.23 | d_m_disc_real: 0.14 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.22 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:41:01,229 (67/67) | batch: 118.7ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:41:34,324 Epoch 4, MPJPE: 88.2255, PA-MPJPE: 54.6182, ACCEL: 27.5513, PVE: 104.2368, ACCEL_ERR: 28.4085, 2020-04-28 17:41:34,466 Epoch 5 performance: 54.6182 2020-04-28 17:42:21,585 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.9196 | loss_kp_2d: 0.66 | loss_kp_3d: 1.19 | loss_shape: 0.01 | loss_pose: 0.36 | e_m_disc_loss: 0.25 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.13 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:42:34,036 (67/67) | batch: 118.8ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:43:06,640 Epoch 5, MPJPE: 88.8532, PA-MPJPE: 56.2239, ACCEL: 27.3844, PVE: 105.3923, ACCEL_ERR: 28.2387, 2020-04-28 17:43:06,768 Epoch 6 performance: 56.2239 2020-04-28 17:43:53,553 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.9362 | loss_kp_2d: 1.73 | loss_kp_3d: 0.83 | loss_shape: 0.01 | loss_pose: 1.63 | e_m_disc_loss: 0.30 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.13 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:44:06,067 (67/67) | batch: 118.3ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:44:38,510 Epoch 6, MPJPE: 88.3176, PA-MPJPE: 55.8464, ACCEL: 26.4748, PVE: 105.4803, ACCEL_ERR: 27.3649, 2020-04-28 17:44:38,713 Epoch 7 performance: 55.8464 2020-04-28 17:45:25,221 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.9353 | loss_kp_2d: 1.00 | loss_kp_3d: 1.12 | loss_shape: 0.02 | loss_pose: 0.75 | e_m_disc_loss: 0.21 | d_m_disc_real: 0.14 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.22 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:45:37,637 (67/67) | batch: 117.2ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:46:10,168 Epoch 7, MPJPE: 88.1955, PA-MPJPE: 55.2046, ACCEL: 26.8387, PVE: 105.0781, ACCEL_ERR: 27.6987, 2020-04-28 17:46:10,294 Epoch 8 performance: 55.2046 2020-04-28 17:46:56,930 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.9401 | loss_kp_2d: 0.94 | loss_kp_3d: 1.26 | e_m_disc_loss: 0.20 | d_m_disc_real: 0.16 | d_m_disc_fake: 0.09 | d_m_disc_loss: 0.25 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:47:09,368 (67/67) | batch: 118.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:47:42,894 Epoch 8, MPJPE: 89.4215, PA-MPJPE: 55.9550, ACCEL: 26.3479, PVE: 106.6388, ACCEL_ERR: 27.2342, 2020-04-28 17:47:43,039 Epoch 9 performance: 55.9550 2020-04-28 17:48:29,652 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.8680 | loss_kp_2d: 0.68 | loss_kp_3d: 1.04 | e_m_disc_loss: 0.20 | d_m_disc_real: 0.16 | d_m_disc_fake: 0.09 | d_m_disc_loss: 0.25 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:48:42,464 (67/67) | batch: 121.5ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 17:49:15,286 Epoch 9, MPJPE: 88.1164, PA-MPJPE: 54.8346, ACCEL: 26.2051, PVE: 103.8222, ACCEL_ERR: 27.1294, 2020-04-28 17:49:15,415 Epoch 10 performance: 54.8346 2020-04-28 17:50:06,119 (500/500) | Total: 0:00:50 | ETA: 0:00:01 | loss: 2.7482 | loss_kp_2d: 0.49 | loss_kp_3d: 0.75 | loss_shape: 0.00 | loss_pose: 0.81 | e_m_disc_loss: 0.23 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.15 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:50:18,572 (67/67) | batch: 118.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:50:51,661 Epoch 10, MPJPE: 86.4797, PA-MPJPE: 55.2742, ACCEL: 25.8963, PVE: 102.9881, ACCEL_ERR: 26.8101, 2020-04-28 17:50:51,805 Epoch 11 performance: 55.2742 2020-04-28 17:51:38,600 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6959 | loss_kp_2d: 0.85 | loss_kp_3d: 1.00 | loss_shape: 0.01 | loss_pose: 0.32 | e_m_disc_loss: 0.24 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.14 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:51:51,210 (67/67) | batch: 120.1ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 17:52:25,249 Epoch 11, MPJPE: 85.8728, PA-MPJPE: 54.6931, ACCEL: 25.5772, PVE: 102.0871, ACCEL_ERR: 26.5127, 2020-04-28 17:52:25,406 Epoch 12 performance: 54.6931 2020-04-28 17:53:12,202 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6467 | loss_kp_2d: 0.83 | loss_kp_3d: 1.09 | loss_shape: 0.03 | loss_pose: 0.58 | e_m_disc_loss: 0.22 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.10 | d_m_disc_loss: 0.15 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:53:24,947 (67/67) | batch: 120.7ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 17:53:58,189 Epoch 12, MPJPE: 86.3866, PA-MPJPE: 55.1251, ACCEL: 25.6505, PVE: 102.5202, ACCEL_ERR: 26.5789, 2020-04-28 17:53:58,334 Epoch 13 performance: 55.1251 2020-04-28 17:54:50,234 (500/500) | Total: 0:00:51 | ETA: 0:00:01 | loss: 2.6951 | loss_kp_2d: 1.10 | loss_kp_3d: 0.71 | loss_shape: 0.01 | loss_pose: 0.37 | e_m_disc_loss: 0.25 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:55:02,641 (67/67) | batch: 118.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:55:35,460 Epoch 13, MPJPE: 86.4423, PA-MPJPE: 55.2600, ACCEL: 25.9393, PVE: 102.5090, ACCEL_ERR: 26.8482, 2020-04-28 17:55:35,645 Epoch 14 performance: 55.2600 2020-04-28 17:56:22,458 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6767 | loss_kp_2d: 1.04 | loss_kp_3d: 0.78 | e_m_disc_loss: 0.24 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.09 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:56:34,586 (67/67) | batch: 116.9ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:57:07,620 Epoch 14, MPJPE: 86.3996, PA-MPJPE: 54.9225, ACCEL: 25.5423, PVE: 102.4451, ACCEL_ERR: 26.4808, 2020-04-28 17:57:07,770 Epoch 15 performance: 54.9225 2020-04-28 17:57:54,521 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7793 | loss_kp_2d: 0.73 | loss_kp_3d: 1.50 | e_m_disc_loss: 0.27 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.12 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:58:06,833 (67/67) | batch: 118.2ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 17:58:39,767 Epoch 15, MPJPE: 86.4569, PA-MPJPE: 54.6776, ACCEL: 25.5324, PVE: 102.3305, ACCEL_ERR: 26.4675, 2020-04-28 17:58:39,950 Epoch 16 performance: 54.6776 2020-04-28 17:59:26,793 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7049 | loss_kp_2d: 0.82 | loss_kp_3d: 0.90 | loss_shape: 0.01 | loss_pose: 0.68 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.03 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 17:59:39,277 (67/67) | batch: 119.4ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:00:11,704 Epoch 16, MPJPE: 86.5707, PA-MPJPE: 54.8733, ACCEL: 25.4781, PVE: 102.3858, ACCEL_ERR: 26.4126, 2020-04-28 18:00:11,883 Epoch 17 performance: 54.8733 2020-04-28 18:00:58,426 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6740 | loss_kp_2d: 1.11 | loss_kp_3d: 0.86 | e_m_disc_loss: 0.36 | d_m_disc_real: 0.09 | d_m_disc_fake: 0.04 | d_m_disc_loss: 0.14 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:01:10,849 (67/67) | batch: 118.1ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:01:43,817 Epoch 17, MPJPE: 86.5733, PA-MPJPE: 54.7636, ACCEL: 25.4780, PVE: 102.4244, ACCEL_ERR: 26.4129, 2020-04-28 18:01:43,945 Epoch 18 performance: 54.7636 2020-04-28 18:02:30,809 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6949 | loss_kp_2d: 0.70 | loss_kp_3d: 0.76 | loss_shape: 0.00 | loss_pose: 0.81 | e_m_disc_loss: 0.36 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:02:43,000 (67/67) | batch: 117.5ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:03:15,695 Epoch 18, MPJPE: 86.5603, PA-MPJPE: 54.7283, ACCEL: 25.4308, PVE: 102.4448, ACCEL_ERR: 26.3688, 2020-04-28 18:03:15,845 Epoch 19 performance: 54.7283 2020-04-28 18:04:03,283 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6551 | loss_kp_2d: 0.65 | loss_kp_3d: 0.74 | e_m_disc_loss: 0.29 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:04:15,465 (67/67) | batch: 117.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:04:47,963 Epoch 19, MPJPE: 86.5021, PA-MPJPE: 54.6963, ACCEL: 25.4630, PVE: 102.3905, ACCEL_ERR: 26.3996, 2020-04-28 18:04:48,115 Epoch 20 performance: 54.6963 2020-04-28 18:05:39,558 (500/500) | Total: 0:00:50 | ETA: 0:00:01 | loss: 2.7341 | loss_kp_2d: 1.15 | loss_kp_3d: 0.67 | e_m_disc_loss: 0.29 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.09 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:05:52,395 (67/67) | batch: 122.7ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:06:25,096 Epoch 20, MPJPE: 86.5212, PA-MPJPE: 54.6335, ACCEL: 25.4595, PVE: 102.3775, ACCEL_ERR: 26.3960, 2020-04-28 18:06:25,223 Epoch 21 performance: 54.6335 2020-04-28 18:07:12,379 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7275 | loss_kp_2d: 1.02 | loss_kp_3d: 0.84 | loss_shape: 0.06 | loss_pose: 1.16 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:07:24,547 (67/67) | batch: 117.3ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:07:57,038 Epoch 21, MPJPE: 86.5159, PA-MPJPE: 54.7687, ACCEL: 25.5565, PVE: 102.3806, ACCEL_ERR: 26.4873, 2020-04-28 18:07:57,190 Epoch 22 performance: 54.7687 2020-04-28 18:08:44,248 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6775 | loss_kp_2d: 0.72 | loss_kp_3d: 0.76 | loss_shape: 0.04 | loss_pose: 1.37 | e_m_disc_loss: 0.32 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.15 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:08:56,404 (67/67) | batch: 116.7ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:09:29,021 Epoch 22, MPJPE: 86.4920, PA-MPJPE: 54.7528, ACCEL: 25.5490, PVE: 102.3578, ACCEL_ERR: 26.4807, 2020-04-28 18:09:29,172 Epoch 23 performance: 54.7528 2020-04-28 18:10:15,994 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6667 | loss_kp_2d: 0.70 | loss_kp_3d: 0.88 | e_m_disc_loss: 0.32 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:10:28,157 (67/67) | batch: 116.8ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:11:00,603 Epoch 23, MPJPE: 86.4850, PA-MPJPE: 54.7576, ACCEL: 25.5387, PVE: 102.3518, ACCEL_ERR: 26.4709, 2020-04-28 18:11:00,783 Epoch 24 performance: 54.7576 2020-04-28 18:11:47,945 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6311 | loss_kp_2d: 0.72 | loss_kp_3d: 0.65 | loss_shape: 0.00 | loss_pose: 0.43 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.10 2020-04-28 18:12:00,262 (67/67) | batch: 117.2ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:12:32,859 Epoch 24, MPJPE: 86.4939, PA-MPJPE: 54.7664, ACCEL: 25.5310, PVE: 102.3616, ACCEL_ERR: 26.4637, 2020-04-28 18:12:33,011 Epoch 25 performance: 54.7664 2020-04-28 18:13:20,511 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7219 | loss_kp_2d: 0.74 | loss_kp_3d: 0.47 | loss_shape: 0.00 | loss_pose: 0.28 | e_m_disc_loss: 0.37 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:13:33,380 (67/67) | batch: 121.6ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:14:05,994 Epoch 25, MPJPE: 86.4563, PA-MPJPE: 54.7460, ACCEL: 25.5327, PVE: 102.3399, ACCEL_ERR: 26.4654, 2020-04-28 18:14:06,122 Epoch 26 performance: 54.7460 2020-04-28 18:14:53,012 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6909 | loss_kp_2d: 0.88 | loss_kp_3d: 0.90 | e_m_disc_loss: 0.29 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.13 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:15:05,170 (67/67) | batch: 117.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:15:37,938 Epoch 26, MPJPE: 86.4457, PA-MPJPE: 54.7430, ACCEL: 25.5239, PVE: 102.3337, ACCEL_ERR: 26.4575, 2020-04-28 18:15:38,133 Epoch 27 performance: 54.7430 2020-04-28 18:16:28,684 (500/500) | Total: 0:00:50 | ETA: 0:00:01 | loss: 2.6425 | loss_kp_2d: 0.74 | loss_kp_3d: 0.87 | loss_shape: 0.00 | loss_pose: 0.79 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:16:41,071 (67/67) | batch: 117.6ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:17:14,417 Epoch 27, MPJPE: 86.4483, PA-MPJPE: 54.7389, ACCEL: 25.5125, PVE: 102.3312, ACCEL_ERR: 26.4467, 2020-04-28 18:17:14,549 Epoch 28 performance: 54.7389 2020-04-28 18:18:01,348 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7038 | loss_kp_2d: 0.55 | loss_kp_3d: 0.84 | e_m_disc_loss: 0.34 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.12 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:18:13,642 (67/67) | batch: 118.5ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:18:46,395 Epoch 28, MPJPE: 86.4503, PA-MPJPE: 54.7388, ACCEL: 25.5113, PVE: 102.3329, ACCEL_ERR: 26.4455, 2020-04-28 18:18:46,561 Epoch 29 performance: 54.7388 2020-04-28 18:19:33,977 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7691 | loss_kp_2d: 0.76 | loss_kp_3d: 0.64 | loss_shape: 0.01 | loss_pose: 0.45 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:19:46,518 (67/67) | batch: 119.6ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:20:19,760 Epoch 29, MPJPE: 86.4516, PA-MPJPE: 54.7397, ACCEL: 25.5113, PVE: 102.3342, ACCEL_ERR: 26.4456, 2020-04-28 18:20:19,889 Epoch 30 performance: 54.7397 2020-04-28 18:21:06,507 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7000 | loss_kp_2d: 0.87 | loss_kp_3d: 0.79 | loss_shape: 0.00 | loss_pose: 0.51 | e_m_disc_loss: 0.31 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:21:18,808 (67/67) | batch: 116.9ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:21:51,760 Epoch 30, MPJPE: 86.4529, PA-MPJPE: 54.7399, ACCEL: 25.5112, PVE: 102.3347, ACCEL_ERR: 26.4454, 2020-04-28 18:21:51,909 Epoch 31 performance: 54.7399 2020-04-28 18:22:43,510 (500/500) | Total: 0:00:51 | ETA: 0:00:01 | loss: 2.7707 | loss_kp_2d: 1.01 | loss_kp_3d: 1.07 | loss_shape: 0.01 | loss_pose: 0.31 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.12 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:22:56,218 (67/67) | batch: 120.1ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:23:29,423 Epoch 31, MPJPE: 86.4545, PA-MPJPE: 54.7412, ACCEL: 25.5098, PVE: 102.3357, ACCEL_ERR: 26.4442, 2020-04-28 18:23:29,553 Epoch 32 performance: 54.7412 2020-04-28 18:24:16,216 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7585 | loss_kp_2d: 0.95 | loss_kp_3d: 0.80 | loss_shape: 0.01 | loss_pose: 0.79 | e_m_disc_loss: 0.27 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.12 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:24:28,556 (67/67) | batch: 117.6ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:25:01,837 Epoch 32, MPJPE: 86.4553, PA-MPJPE: 54.7427, ACCEL: 25.5109, PVE: 102.3368, ACCEL_ERR: 26.4452, 2020-04-28 18:25:01,980 Epoch 33 performance: 54.7427 2020-04-28 18:25:48,806 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6797 | loss_kp_2d: 0.62 | loss_kp_3d: 0.64 | loss_shape: 0.01 | loss_pose: 0.44 | e_m_disc_loss: 0.27 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.14 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:26:01,470 (67/67) | batch: 120.1ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:26:34,403 Epoch 33, MPJPE: 86.4581, PA-MPJPE: 54.7434, ACCEL: 25.5093, PVE: 102.3390, ACCEL_ERR: 26.4437, 2020-04-28 18:26:34,562 Epoch 34 performance: 54.7434 2020-04-28 18:27:21,259 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7101 | loss_kp_2d: 0.57 | loss_kp_3d: 0.60 | loss_shape: 0.01 | loss_pose: 0.76 | e_m_disc_loss: 0.26 | d_m_disc_real: 0.09 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:27:33,998 (67/67) | batch: 120.8ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:28:07,159 Epoch 34, MPJPE: 86.4568, PA-MPJPE: 54.7435, ACCEL: 25.5098, PVE: 102.3383, ACCEL_ERR: 26.4441, 2020-04-28 18:28:07,290 Epoch 35 performance: 54.7435 2020-04-28 18:28:53,883 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6428 | loss_kp_2d: 0.86 | loss_kp_3d: 0.59 | e_m_disc_loss: 0.30 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:29:06,128 (67/67) | batch: 116.3ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:29:39,133 Epoch 35, MPJPE: 86.4554, PA-MPJPE: 54.7427, ACCEL: 25.5101, PVE: 102.3382, ACCEL_ERR: 26.4445, 2020-04-28 18:29:39,281 Epoch 36 performance: 54.7427 2020-04-28 18:30:26,179 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7171 | loss_kp_2d: 1.22 | loss_kp_3d: 1.02 | loss_shape: 0.02 | loss_pose: 0.58 | e_m_disc_loss: 0.40 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.03 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:30:38,404 (67/67) | batch: 117.2ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:31:11,132 Epoch 36, MPJPE: 86.4577, PA-MPJPE: 54.7441, ACCEL: 25.5088, PVE: 102.3392, ACCEL_ERR: 26.4432, 2020-04-28 18:31:11,318 Epoch 37 performance: 54.7441 2020-04-28 18:31:58,223 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.5947 | loss_kp_2d: 1.28 | loss_kp_3d: 0.95 | loss_shape: 0.02 | loss_pose: 0.94 | e_m_disc_loss: 0.34 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:32:10,617 (67/67) | batch: 117.9ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:32:43,657 Epoch 37, MPJPE: 86.4566, PA-MPJPE: 54.7438, ACCEL: 25.5082, PVE: 102.3384, ACCEL_ERR: 26.4426, 2020-04-28 18:32:43,811 Epoch 38 performance: 54.7438 2020-04-28 18:33:31,157 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6832 | loss_kp_2d: 0.49 | loss_kp_3d: 0.70 | loss_shape: 0.02 | loss_pose: 1.42 | e_m_disc_loss: 0.27 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:33:44,135 (67/67) | batch: 123.5ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:34:17,023 Epoch 38, MPJPE: 86.4575, PA-MPJPE: 54.7450, ACCEL: 25.5070, PVE: 102.3389, ACCEL_ERR: 26.4415, 2020-04-28 18:34:17,169 Epoch 39 performance: 54.7450 2020-04-28 18:35:03,834 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6798 | loss_kp_2d: 0.99 | loss_kp_3d: 0.76 | loss_shape: 0.01 | loss_pose: 0.50 | e_m_disc_loss: 0.30 | d_m_disc_real: 0.09 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.16 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:35:16,499 (67/67) | batch: 120.4ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:35:49,399 Epoch 39, MPJPE: 86.4579, PA-MPJPE: 54.7455, ACCEL: 25.5073, PVE: 102.3389, ACCEL_ERR: 26.4418, 2020-04-28 18:35:49,545 Epoch 40 performance: 54.7455 2020-04-28 18:36:37,016 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6764 | loss_kp_2d: 0.83 | loss_kp_3d: 0.64 | loss_shape: 0.02 | loss_pose: 0.74 | e_m_disc_loss: 0.28 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.13 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.10 2020-04-28 18:36:49,772 (67/67) | batch: 120.7ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:37:22,523 Epoch 40, MPJPE: 86.4577, PA-MPJPE: 54.7448, ACCEL: 25.5076, PVE: 102.3405, ACCEL_ERR: 26.4421, 2020-04-28 18:37:22,670 Epoch 41 performance: 54.7448 2020-04-28 18:38:19,826 (500/500) | Total: 0:00:56 | ETA: 0:00:01 | loss: 2.6005 | loss_kp_2d: 1.05 | loss_kp_3d: 0.59 | loss_shape: 0.01 | loss_pose: 0.57 | e_m_disc_loss: 0.36 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:38:32,518 (67/67) | batch: 119.9ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:39:05,219 Epoch 41, MPJPE: 86.4564, PA-MPJPE: 54.7423, ACCEL: 25.5072, PVE: 102.3389, ACCEL_ERR: 26.4417, 2020-04-28 18:39:05,347 Epoch 42 performance: 54.7423 2020-04-28 18:39:52,108 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7966 | loss_kp_2d: 0.80 | loss_kp_3d: 0.59 | loss_shape: 0.01 | loss_pose: 0.56 | e_m_disc_loss: 0.33 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:40:04,600 (67/67) | batch: 119.1ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:40:37,272 Epoch 42, MPJPE: 86.4580, PA-MPJPE: 54.7440, ACCEL: 25.5074, PVE: 102.3408, ACCEL_ERR: 26.4419, 2020-04-28 18:40:37,399 Epoch 43 performance: 54.7440 2020-04-28 18:41:24,532 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7288 | loss_kp_2d: 1.10 | loss_kp_3d: 2.34 | e_m_disc_loss: 0.28 | d_m_disc_real: 0.03 | d_m_disc_fake: 0.07 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:41:37,029 (67/67) | batch: 118.4ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:42:10,256 Epoch 43, MPJPE: 86.4579, PA-MPJPE: 54.7424, ACCEL: 25.5061, PVE: 102.3399, ACCEL_ERR: 26.4407, 2020-04-28 18:42:10,401 Epoch 44 performance: 54.7424 2020-04-28 18:42:57,285 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6371 | loss_kp_2d: 0.73 | loss_kp_3d: 0.86 | loss_shape: 0.01 | loss_pose: 0.99 | e_m_disc_loss: 0.34 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.10 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:43:10,172 (67/67) | batch: 121.7ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:43:43,008 Epoch 44, MPJPE: 86.4590, PA-MPJPE: 54.7436, ACCEL: 25.5054, PVE: 102.3400, ACCEL_ERR: 26.4400, 2020-04-28 18:43:43,136 Epoch 45 performance: 54.7436 2020-04-28 18:44:30,157 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.7158 | loss_kp_2d: 0.64 | loss_kp_3d: 0.77 | loss_shape: 0.01 | loss_pose: 1.02 | e_m_disc_loss: 0.31 | d_m_disc_real: 0.04 | d_m_disc_fake: 0.08 | d_m_disc_loss: 0.12 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:44:42,585 (67/67) | batch: 120.0ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:45:15,383 Epoch 45, MPJPE: 86.4590, PA-MPJPE: 54.7435, ACCEL: 25.5045, PVE: 102.3402, ACCEL_ERR: 26.4392, 2020-04-28 18:45:15,546 Epoch 46 performance: 54.7435 2020-04-28 18:46:02,232 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6847 | loss_kp_2d: 0.74 | loss_kp_3d: 0.58 | loss_shape: 0.01 | loss_pose: 0.76 | e_m_disc_loss: 0.31 | d_m_disc_real: 0.06 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.11 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:46:14,368 (67/67) | batch: 116.4ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:46:47,041 Epoch 46, MPJPE: 86.4583, PA-MPJPE: 54.7435, ACCEL: 25.5045, PVE: 102.3402, ACCEL_ERR: 26.4391, 2020-04-28 18:46:47,202 Epoch 47 performance: 54.7435 2020-04-28 18:47:34,374 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6367 | loss_kp_2d: 0.92 | loss_kp_3d: 0.58 | loss_shape: 0.00 | loss_pose: 0.33 | e_m_disc_loss: 0.32 | d_m_disc_real: 0.05 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.09 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:47:47,335 (67/67) | batch: 122.2ms | Total: 0:00:12 | ETA: 0:00:01 2020-04-28 18:48:20,398 Epoch 47, MPJPE: 86.4597, PA-MPJPE: 54.7439, ACCEL: 25.5030, PVE: 102.3411, ACCEL_ERR: 26.4377, 2020-04-28 18:48:20,595 Epoch 48 performance: 54.7439 2020-04-28 18:49:07,432 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6562 | loss_kp_2d: 0.91 | loss_kp_3d: 0.94 | loss_shape: 0.00 | loss_pose: 1.43 | e_m_disc_loss: 0.31 | d_m_disc_real: 0.07 | d_m_disc_fake: 0.06 | d_m_disc_loss: 0.13 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:49:19,865 (67/67) | batch: 118.2ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:49:52,430 Epoch 48, MPJPE: 86.4590, PA-MPJPE: 54.7430, ACCEL: 25.5033, PVE: 102.3402, ACCEL_ERR: 26.4380, 2020-04-28 18:49:52,557 Epoch 49 performance: 54.7430 2020-04-28 18:50:39,425 (500/500) | Total: 0:00:46 | ETA: 0:00:01 | loss: 2.6129 | loss_kp_2d: 1.01 | loss_kp_3d: 0.71 | loss_shape: 0.01 | loss_pose: 0.44 | e_m_disc_loss: 0.38 | d_m_disc_real: 0.08 | d_m_disc_fake: 0.05 | d_m_disc_loss: 0.13 | data: 0.01 | forward: 0.02 | loss: 0.01 | backward: 0.06 | batch: 0.09 2020-04-28 18:50:51,909 (67/67) | batch: 118.6ms | Total: 0:00:11 | ETA: 0:00:01 2020-04-28 18:51:25,130 Epoch 49, MPJPE: 86.4603, PA-MPJPE: 54.7443, ACCEL: 25.5021, PVE: 102.3413, ACCEL_ERR: 26.4369, 2020-04-28 18:51:25,271 Epoch 50 performance: 54.7443

When I disable those 2D datasets, I got eval result like 54.0, which still have 2mm gap with the paper. And I find that the eval error increases, as the training goes. This situation is common?

@athn-nik Can you provide the preprocessed Kinetics dataset or preprocessing script?

@Frank-Dz Hi did you pre-process the AMASS and 3DHP dataset? If yes than kindly share with us. Facing problem in pre-processing of these two datasets especially in AMASS CMU part.

@Frank-Dz Hi did you pre-process the AMASS and 3DHP dataset? If yes than kindly share with us. Facing problem in pre-processing of these two datasets especially in AMASS CMU part.

Hi~ Sorry. Currently, I am quite busy with another project. Maybe I will check and run the code a long time later. Once I make it I will let you know.

I'm confused about AMASS dataset... On AMASS there are different databases, from below which database we need? Furthermore we need only body data?

No, AMASS in a dataset that unifies all those data. It has its own website where you can find the data and download them.

@athn-nik I'm following same website that you suggested, In this website I found above databases mentioned in screenshot. Kindly have a look again and help us to find right dataset.

All of them. Amass is the union of all those MoCap data to a single dataset in SMPL/SMPL-H format. Please refer to the original dataset for more details. Yes, you need the body data. Body pose parameters.

@athn-nik Hi, sorry for bothering you, but I can't get to the download page as in the pic you provided with that url: https://amass.is.tue.mpg.de/ it points to the main site of AMASS dataset, and it looks like this in my explorer: Screenshot from 2020-05-17 21-07-40

and if I click on the download button, it shows: Screenshot from 2020-05-17 21-09-03

every url in this site directly lead me to downloading a file. but obviously there're less than 15 files, and I just wonder why you guys can see this page: Annotation 2020-04-28 180923

maybe because AMASS edited their download site? do they still provide the whole AMASS dataset that is described in their paper? (or I just missed sth?) besides today is 5.17.2020

sorry, I found that page, the dataset button :-( https://amass.is.tue.mpg.de/dataset my fault :-(

Hello! I run the training code with the default config file you provided on the full training dataset . But when I run the test code on the 3DPW dataset, I got the different results when PA-MPJPE is 57.2mm while result in the paper is 51.9mm. Do I miss something?

I got difficuly on getting PennAction. The download link provided by the offical website http://dreamdragon.github.io/PennAction/ is invaild. I would appreciate it if you kindly tell me how you get PennAction.

@lixincheng4026 try this link https://upenn.app.box.com/v/PennAction

@lan786 Thanks.It turns out that i was block by the firewalls. I download this file, but when I tried to unzip this tar.gz, there was another error. The file I download has 3G, but the file after unzipped has only 1.5MB. Is there something wrong with the way I unzipped it? I tried "tar xzvf" and unzipped in Windows

@lixincheng4026 I'm not sure about this problem. May be you can try in Ubuntu but I think windows is also ok

Hi, I meet the same question. I run the training code with the default config file (exclude 2D dataset PoseTrack and PennAction ), and can only get PA-MPJPE with 53.5mm on the 3DPW test dataset. By the way, the pre-processing details of Human 3.6m dataset is not given and I can't reproduce the same result on it.

@ZephirGe @Ling-wei Hi，I have the same question,when I use dataset Insta\ThreeDPW\MPII3D,with the default cfg,i get a result 55mm.When i try other dataset combination,i can get best result is 54.2mm. If you have reproduce the similar result with paper,please tell me,thx! by the way,i try use my best model to test the demo,i found the result is bader than the official model.

why not training together with PoseTrack and PennAction?

I have the same problem, using the provided config file I am not able to reproduce the results published in the paper. I also repeated the experiment for 20 times to exclude any random effects.

I would highly appreciate it if you could update the repository accordingly such that the results are reproducible.

Dear all,

I double-checked everything to see if I can reproduce the results. Everything seems correct and I could reproduce the results reported in the paper. I committed the config files I used, see: https://github.com/mkocabas/VIBE/commit/cee26151ea880ea7547e78afd7faff92b102a353. Besides, I got several emails from people who can reproduce the results. So, I am not the only one who can achieve this.

There are a couple of things we need to be careful when interpreting the results. The results we see on the terminal output or train_log.txt file obtained at the end of each epoch are on the validation set. For the 3DPW dataset, validation set is harder than the test set, so it is expected to get worse results. You need to run eval.py script to get the results on the test set.

PennAction and PoseTrack are quite small datasets in comparison to InstaVariety. So, we couldn't observe any improvements using them. I would suggest using InstaVariety for now.

Thanks!

Thanks for the quick response! I did not check yet, but the usage of the train vs validation set in 3DPW probably explains the difference in performance.

Just as a small note, in your config D_MOTION_LOSS_W is set to 0.5 while in the supp. material of the paper it is reported as 2.0. But I guess this won't make a significant difference.

Hi @mkocabas, really thanks for your work! Could you share the code of how to process human 3.6M and how to set its related hyper-parameters in training?

Seems like the issue is resolved. Feel free to post if you have additional questions.

hey, i want to predict smpl h parameters using vibe. any ideas on how can i achieve this?

Hi, @mkocabas . I meet the same question. When I use dataset Insta\ThreeDPW\MPII3D with the other default configurations, I only get the results: "PA-MPJPE: 54.9796, ACCEL: 27.6044 ", which are much higher than the figures in your paper.

Hi @tangjiapeng,

You can rely on the numbers that you are able to reproduce for now. There is a mistake in experiments using 3DPW during training, you can refer to this https://github.com/mkocabas/VIBE/issues/99#issuecomment-654309245 for a detailed explanation. I am about to finalize the correction, will update the results soon. Sorry for the inconvenience.

Okay, thanks for your quick response!

@ZephirGe My result looks similar to author's !! How about you?

@tangjiapeng hello, how did u train with 3dpw dataset, the prepared-train-dateset didnt include 3dpw for train?

@tangjiapeng hello, how did u train with 3dpw dataset, the prepared-train-dateset didnt include 3dpw for train?

This is because pursuant to this issue, the authors removed the code which generated 3dpw training data. This is because there was a data leak and moreover once fixed inclusion of 3dpw training data hurt qualitative results.

At least, that's what I have been able to gather after pouring over the Issues.

mkocabas / VIBE

About the training process #69