alexsnow348 / facetag

Face Recognition system trained for 81 people. Accuracy is around 70% achieved.

FaceNet Implementation:

Face Recognition system trained using FaceNet for 81 people. Accuracy is around 70% achieved.

Details Explanation of FaceNet: here

Click here for Details Explanation in Burmese YouTube Video

Real-time test result:

Steps on how to run:

1. Download miniconda3 and install.

2. Install conda evn and required pip library packages and go the environment.

conda env create -f environment.yml
pip install -r requirement.txt
conda activate CV

3. (Optional) If want to train on own image, follow the instructions from Facenet.

4. Run the test on single image.

python image.py --img_filename [PATH OF TEST IMAGE] --save_filename [NAME TO SAVE RESULT]

5. Run the real time video.

python video.py --video_link [0(WEB CAM) or 1(USB CAM) ]