OpenVINO Training Extensions - Real-time 2D Multi-Person Pose Estimation

[junxnone]

Reference

• [ ] Real-time 2D Multi-Person Pose Estimation
• [ ] 论文- Real-time 2D Multi-Person Pose Estimation on CPU: Lightweight OpenPose

Setup Environment

bash init_venv.sh

Download pre-trained MobileNet v1 weights

Baidu Netdisk - mobilenet_sgd_68.848.pth.tar

 python scripts/prepare_train_labels.py \
--labels  mscoco2017/annotations/person_keypoints_train2017.json
python scripts/make_val_subset.py \
--labels mscoco2017/annotations/person_keypoints_val2017.json

Train

ulimit -n 65536
export CUDA_VISIBLE_DEVICES=1,2
python train.py --train-images-folder  mscoco2017/train2017/ \
--prepared-train-labels prepared_train_annotation.pkl \
--val-labels val_subset.json --val-images-folder mscoco2017/val2017/ \
--checkpoint-path mobilenet_sgd_68.848.pth.tar --weights-only

太多GPU 会引发 ulimit issue

.....
Iter: 45000
stage1_pafs_loss:     111.90049415588379
stage1_heatmaps_loss: 35.61218692779541
stage2_pafs_loss:     96.17975372314453
stage2_heatmaps_loss: 29.904519863128662
Validation...
Running test for keypoints results.
loading annotations into memory...
Done (t=0.01s)
creating index...
index created!
Loading and preparing results...
DONE (t=0.02s)
creating index...
index created!
Running per image evaluation...
Evaluate annotation type *keypoints*
DONE (t=0.22s).
Accumulating evaluation results...
DONE (t=0.00s).
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets= 20 ] = 0.322
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets= 20 ] = 0.554
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets= 20 ] = 0.323
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets= 20 ] = 0.291
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets= 20 ] = 0.382
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 20 ] = 0.368
 Average Recall     (AR) @[ IoU=0.50      | area=   all | maxDets= 20 ] = 0.596
 Average Recall     (AR) @[ IoU=0.75      | area=   all | maxDets= 20 ] = 0.374
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets= 20 ] = 0.301
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets= 20 ] = 0.456

Validation

python val.py \
--labels mscoco2017/annotations/person_keypoints_val2017.json \
--images-folder mscoco2017/val2017 \
--checkpoint-path default_checkpoints/checkpoint_iter_245000.pth.tar \
--multiscale

[junxnone]

79