human_detection Environment Building

This project is building from tf-pose-estimation, fsanet-pytorch and webcam in the following link:

• GitHub: ZheC/tf-pose-estimation
• GitHub: omasaht/headpose-fsanet-pytorch
• GitHub: jhukaby/webcamjs

After Install environment

• Download the pretrained model from the link of tf-pose-estimation and headpose-fsanet-pytorch and put it in
  • src/services/detection/fsanet_pytorch/pretrained
  • src/services/detection/tf_pose/models/graph
  • src/services/detection/tf_pose/models/pretrained
• Install python3, python3-pip, virtualenv with pip
• Edit ".env" file.
  • hasGPU: true if GPU exist, 'None' if GPU doesn't exist
  • setup HOST with the host ip

Steps:

1. Build environment with virtualenv.

   $ make BuildENV
   

2. Activate virtualenv.

   $ source bin/activate
   

3. Install dependency packages.

   $ make InstallPackage
   

4. Run test

   $ make TEST
   

5. Run API server

   $ make run
   

   API config is 'config.py'.

   There are 4 models for tf_pose can be selected in ModelConfig.TF_POSE_TYPE.