Uses Python3 + Tensorflow 1.4
- Tensorflow
- Python 3
- Numpy
- OpenCV (3.4.X to support RTSP)
- Pillow (pip install Pillow)
- dLib (IMPORTANT: install from Source only)
- face_recognition (pip install face_recognition)
-
Dowload the pretrained model from here and place in the model directory in the root of this project
-
Create a directory POI in the root of this project and place images of the person that you wish to track (accepted formats are *.jpg, *.png)
python inference_video_face.py
- --mode (Video / Camera / IP)
- --file_path (Optional : Video File Path. Required if mode = Video only)
- --uname (Optional: Requirded only if mode is IP for streaming from IP camera. Username for RTSP)
- --secret (Optional: Requirded only if mode is IP for streaming from IP camera. Password for RTSP)
- --addr (Optional: Requirded only if mode is IP for streaming from IP camera. ip address of camera