The project detects faces, emotions and predicting gender from faces in a video using the DeepFace library.
The video is processed frame by frame, with each detected face outlined with a green rectangle. The dominant emotion and predicted gender are annotated on the video in lilac text.
- Face Detection: Detects faces in each frame of the video.
- Emotion Detection: Identifies the dominant emotion for each detected face.
- Gender Prediction: Predicts the gender of each detected face.
Mixed success:
- The main face is detected properly when there is a single person in the frame, however, there are small "ghosts" around (funny how men's ties are detected as faces).
- When there are a bunch of people in the frame, the gender and emotions of each person are detected with low accuracy.
Exciting to see the parliamentary debate video packed with emotions!
