my test with EfficientPose in Pytorch

Demo

# install requirements
pip install pycocotools numpy opencv-python tqdm tensorboard pyyaml webcolors
pip install torch==1.12.1+cu113
pip install torchvision==0.13.0+cu113
 
# run the simple inference script
python inference.py

Training

Train a custom dataset with pretrained weights (Highly Recommended)

python ./train_pose_WMSA.py --weights ./weights/trained_WMSA/obj_8/efficientpose-d0_linemod_obj8_one_best_train.pth  --lr  1e-3  --batch_size 2

4. Early stopping a training session

# while training, press Ctrl+c, the program will catch KeyboardInterrupt
# and stop training, save current checkpoint.

6. Evaluate model performance

# eval on your_project, efficientdet-d5
python evaluate_WMSA.py --object_id 8  --weights  ./weights/trained_WMSA/obj_8/efficientpose-d0_linemod_obj8_one_best_train.pth

References

Appreciate the great work from the following repositories:

• google/automl
• lukemelas/EfficientNet-PyTorch
• signatrix/efficientdet
• vacancy/Synchronized-BatchNorm-PyTorch
• https://github.com/claire-s11
• https://github.com/zylo117/Yet-Another-EfficientDet-Pytorch