search

nizhenliang / CGRSeg

[ECCV 2024] Context-Guided Spatial Feature Reconstruction for Efficient Semantic Segmentation
Apache License 2.0
55 stars 3 forks source link

problem about training #12

Open dongxiaofei12 opened 1 week ago

dongxiaofei12 commented 1 week ago

image When I started training, a large number of missing keys as shown in the image above appeared. Is this normal?The mmsegmentation I used was version 1.2.2

dongxiaofei12 commented 1 week ago

1728904931295 Is there an issue with the new version?

nizhenliang commented 1 week ago

How many cards did you use? What is the batch size? Please provide more training parameter settings.

nizhenliang commented 1 week ago

The batch size we used is 16. In addition, the pre-trained model of the backbone should be loaded.

dongxiaofei12 commented 1 week ago

How many cards did you use? What is the batch size? Please provide more training parameter settings.

2024/10/15 13:48:48 - mmengine - INFO -

System environment: sys.platform: linux Python: 3.8.18 | packaged by conda-forge | (default, Oct 10 2023, 15:44:36) [GCC 12.3.0] CUDA available: True MUSA available: False numpy_random_seed: 0 GPU 0: NVIDIA GeForce RTX 3080 Ti CUDA_HOME: /usr NVCC: Cuda compilation tools, release 9.1, V9.1.8 GCC: gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0 PyTorch: 1.8.0+cu111 PyTorch compiling details: PyTorch built with:

Runtime environment: cudnn_benchmark: True mp_cfg: {'mp_start_method': 'fork', 'opencv_num_threads': 0} dist_cfg: {'backend': 'nccl'} seed: 0 Distributed launcher: none Distributed training: False GPU number: 1

2024/10/15 13:48:48 - mmengine - INFO - Config: EfficientFormer_depth = dict( L=( 5, 5, 15, 10, ), S0=( 2, 2, 6, 4, ), S1=( 3, 3, 9, 6, ), S2=( 4, 4, 12, 8, )) EfficientFormer_expansion_ratios = dict( L=( 4, 4, ( 4, 4, 4, 4, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, ), ( 4, 4, 4, 3, 3, 3, 3, 4, 4, 4, ), ), S0=( 4, 4, ( 4, 3, 3, 3, 4, 4, ), ( 4, 3, 3, 4, ), ), S1=( 4, 4, ( 4, 4, 3, 3, 3, 3, 4, 4, 4, ), ( 4, 4, 3, 3, 4, 4, ), ), S2=( 4, 4, ( 4, 4, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, ), ( 4, 4, 3, 3, 3, 3, 4, 4, ), )) EfficientFormer_width = dict( L=( 40, 80, 192, 384, ), S0=( 32, 48, 96, 176, ), S1=( 32, 48, 120, 224, ), S2=( 32, 64, 144, 288, )) crop_size = ( 256, 256, ) data = dict(samples_per_gpu=4, workers_per_gpu=8) data_preprocessor = dict( bgr_to_rgb=True, mean=[ 123.675, 116.28, 103.53, ], pad_val=0, seg_pad_val=255, size=( 256, 256, ), std=[ 58.395, 57.12, 57.375, ], type='SegDataPreProcessor') data_root = '../data/custom-tif-256' dataset_type = 'PascalVOCDataset' default_hooks = dict( checkpoint=dict(by_epoch=False, interval=16000, type='CheckpointHook'), logger=dict(interval=50, log_metric_by_epoch=False, type='LoggerHook'), param_scheduler=dict(type='ParamSchedulerHook'), sampler_seed=dict(type='DistSamplerSeedHook'), timer=dict(type='IterTimerHook'), visualization=dict(type='SegVisualizationHook')) default_scope = 'mmseg' env_cfg = dict( cudnn_benchmark=True, dist_cfg=dict(backend='nccl'), mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0)) evaluation = dict(interval=4000) find_unused_parameters = True img_ratios = [ 0.5, 0.75, 1.0, 1.25, 1.5, 1.75, ] launcher = 'none' load_from = None log_level = 'INFO' log_processor = dict(by_epoch=False) lr_config = dict( delete=True, by_epoch=False, min_lr=0.0, policy='poly', power=1.0, warmup='linear', warmup_iters=1500, warmup_ratio=1e-06) model = dict( backbone=dict( init_cfg=dict( checkpoint= '/home/badwater/.cache/torch/hub/checkpoints/eformer_s2_450.pth', type='Pretrained'), style='pytorch', type='efficientformerv2_s2_feat'), data_preprocessor=dict( bgr_to_rgb=True, mean=[ 123.675, 116.28, 103.53, ], pad_val=0, seg_pad_val=255, size=( 256, 256, ), std=[ 58.395, 57.12, 57.375, ], type='SegDataPreProcessor'), decode_head=dict( align_corners=False, channels=256, dropout_ratio=0.1, dw_size=11, in_channels=[ 64, 144, 288, ], in_index=[ 1, 2, 3, ], is_dw=True, loss_decode=dict( loss_weight=1.0, type='CrossEntropyLoss', use_sigmoid=False), module='RCA', neck_size=11, next_repeat=6, norm_cfg=dict(requires_grad=True, type='SyncBN'), num_classes=5, ratio=1, square_kernel_size=3, type='CGRSeg'), test_cfg=dict(mode='whole'), train_cfg=dict(), type='EncoderDecoder') norm_cfg = dict(requires_grad=True, type='SyncBN') optim_wrapper = dict( clip_grad=None, optimizer=dict(lr=0.01, momentum=0.9, type='SGD', weight_decay=0.0005), type='OptimWrapper') optimizer = dict( betas=( 0.9, 0.999, ), lr=0.00012, paramwise_cfg=dict( custom_keys=dict( absolute_pos_embed=dict(decay_mult=0.0), head=dict(lr_mult=10.0), norm=dict(decay_mult=0.0))), type='AdamW', weight_decay=0.01) param_scheduler = [ dict( begin=0, by_epoch=False, end=160000, eta_min=0.0001, power=0.9, type='PolyLR'), ] randomness = dict(seed=0) resume = False test_cfg = dict(type='TestLoop') test_dataloader = dict( batch_size=1, dataset=dict( ann_file='ImageSets/Segmentation/test.txt', data_prefix=dict( img_path='TIFFImages', seg_map_path='SegmentationClassPNG'), data_root='../data/custom-tif-256', pipeline=[ dict(type='LoadImageFromFile'), dict(keep_ratio=True, scale=( 2048, 512, ), type='Resize'), dict(type='LoadAnnotations'), dict(type='PackSegInputs'), ], type='PascalVOCDataset'), num_workers=4, persistent_workers=True, sampler=dict(shuffle=False, type='DefaultSampler')) test_evaluator = dict( iou_metrics=[ 'mIoU', ], type='IoUMetric') test_pipeline = [ dict(type='LoadImageFromFile'), dict(keep_ratio=True, scale=( 2048, 512, ), type='Resize'), dict(type='LoadAnnotations'), dict(type='PackSegInputs'), ] train_cfg = dict( max_iters=160000, type='IterBasedTrainLoop', val_interval=16000) train_dataloader = dict( batch_size=4, dataset=dict( ann_file='ImageSets/Segmentation/train.txt', data_prefix=dict( img_path='TIFFImages', seg_map_path='SegmentationClassPNG'), data_root='../data/custom-tif-256', pipeline=[ dict(type='LoadImageFromFile'), dict(type='LoadAnnotations'), dict( keep_ratio=True, ratio_range=( 0.5, 2.0, ), scale=( 2048, 512, ), type='RandomResize'), dict( cat_max_ratio=0.75, crop_size=( 256, 256, ), type='RandomCrop'), dict(prob=0.5, type='RandomFlip'), dict(type='PhotoMetricDistortion'), dict(type='PackSegInputs'), ], type='PascalVOCDataset'), num_workers=4, persistent_workers=True, sampler=dict(shuffle=True, type='DefaultSampler')) train_pipeline = [ dict(type='LoadImageFromFile'), dict(type='LoadAnnotations'), dict( keep_ratio=True, ratio_range=( 0.5, 2.0, ), scale=( 2048, 512, ), type='RandomResize'), dict(cat_max_ratio=0.75, crop_size=( 256, 256, ), type='RandomCrop'), dict(prob=0.5, type='RandomFlip'), dict(type='PhotoMetricDistortion'), dict(type='PackSegInputs'), ] tta_model = dict(type='SegTTAModel') tta_pipeline = [ dict(backend_args=None, type='LoadImageFromFile'), dict( transforms=[ [ dict(keep_ratio=True, scale_factor=0.5, type='Resize'), dict(keep_ratio=True, scale_factor=0.75, type='Resize'), dict(keep_ratio=True, scale_factor=1.0, type='Resize'), dict(keep_ratio=True, scale_factor=1.25, type='Resize'), dict(keep_ratio=True, scale_factor=1.5, type='Resize'), dict(keep_ratio=True, scale_factor=1.75, type='Resize'), ], [ dict(direction='horizontal', prob=0.0, type='RandomFlip'), dict(direction='horizontal', prob=1.0, type='RandomFlip'), ], [ dict(type='LoadAnnotations'), ], [ dict(type='PackSegInputs'), ], ], type='TestTimeAug'), ] val_cfg = dict(type='ValLoop') val_dataloader = dict( batch_size=1, dataset=dict( ann_file='ImageSets/Segmentation/val.txt', data_prefix=dict( img_path='TIFFImages', seg_map_path='SegmentationClassPNG'), data_root='../data/custom-tif-256', pipeline=[ dict(type='LoadImageFromFile'), dict(keep_ratio=True, scale=( 2048, 512, ), type='Resize'), dict(type='LoadAnnotations'), dict(type='PackSegInputs'), ], type='PascalVOCDataset'), num_workers=4, persistent_workers=True, sampler=dict(shuffle=False, type='DefaultSampler')) val_evaluator = dict( iou_metrics=[ 'mIoU', ], type='IoUMetric') vis_backends = [ dict(type='LocalVisBackend'), dict(type='TensorboardVisBackend'), ] visualizer = dict( name='visualizer', type='SegLocalVisualizer', vis_backends=[ dict(type='LocalVisBackend'), dict(type='TensorboardVisBackend'), ]) work_dir = './work_dirs/cgrseg-b-256×256'

2024/10/15 13:48:54 - mmengine - INFO - Distributed training is not used, all SyncBatchNorm (SyncBN) layers in the model will be automatically reverted to BatchNormXd layers if they are used.

2024/10/15 13:48:54 - mmengine - WARNING - The prefix is not set in metric class IoUMetric. Name of parameter - Initialization information

This is my training log.

nizhenliang commented 1 week ago

You only use 1 gpu，the samples_per_gpu should be 16. The batch size should be 16. Now, you use bs 4.

dongxiaofei12 commented 1 week ago

You only use 1 gpu，the samples_per_gpu should be 16. The batch size should be 16. Now, you use bs 4. I modified the batch_size according to the new version, but the training results are still unsatisfactory.