This pipeline is designed to process a video and execute the following operations:
- Extract frames from the video;
- Find a human's face in each frame if there is one;
- Determine locations of bounding boxes around the face and its parts such as mouth, nose and eyes;
- Determine locations of pupils;
- Determine location of all facial points and draw a facial mesh;
- Put all the frames back into a video file.
Note: at each stage of the pipeline all the data acquired is saved in a database(PostgreSQL in this case).
 |
| Initial Frame |
 |
| Bounding Boxes |
 |
| Pupils |
 |
| Facial Mesh |
No comments:
Post a Comment