Shankar vishnu

Code is already given. You can check [here](https://github.com/shankarvishnu/Facenet-Real-time-face-recognition-using-deep-learning-Tensorflow/blob/master/train_main.py).

Code is updated. You should get clone of updated code from [here](https://github.com/shankarvishnu/Facenet-Real-time-face-recognition-using-deep-learning-Tensorflow/blob/master/train_main.py) and follow its proper instructions.

You have to update [this line](https://github.com/shankarvishnu/Facenet-Real-time-face-recognition-using-deep-learning-Tensorflow/blob/b351d5d3f4766572868236eed78a25082da44c27/identify_face_video.py#L50) to use webcam. video_capture = cv2.VideoCapture(input_video) to video_capture = cv2.VideoCapture(0)

You can use multiprocessing for streaming the live-camera feed. For batter reference you can visit [here](https://www.pyimagesearch.com/2019/04/15/live-video-streaming-over-network-with-opencv-and-imagezmq/).

Follow this [comment](https://github.com/AISangam/Facenet-Real-time-face-recognition-using-deep-learning-Tensorflow/issues/17#issuecomment-632615072) to resolve this issue.

Yes, You can test it at your end properly.

"train_img" contain real-life images and "pre_img" contains process and aligned images. This alignment is process of cropping face and resizing them in same dimension. Also it contain information of face...

You can follow this [comment](https://github.com/AISangam/Facenet-Real-time-face-recognition-using-deep-learning-Tensorflow/issues/26#issuecomment-632611542) to ressolve this issue.