Skip to content

Latest commit

 

History

History
26 lines (20 loc) · 1.04 KB

README.md

File metadata and controls

26 lines (20 loc) · 1.04 KB

Face-analysis

Face detection, recognition, gender classification, emotion classification

Alt Text

Features:

  • multiple Face Detectors
  • Face Recognition
  • faciallandmarks detection
  • race Classification
  • gender Classification
  • emotion Detection ['angry', 'disgust', 'fear', 'happy', 'sad', 'surprise', 'neutral']

Notes:

the process of analysis is kinda slow because i use seperate models in this pipeline and it needs some optimizations like:

  • models quantization and using tensorRT For speeding the models
  • most of the faces models (age, gender, race, emotions) is using the same backbone and it can be groubed into one model that need to be retrained again with available datasets.
  • recognition process is applyed for each frame (used as face tracker) and it can be replaced with a suitable object tracker.

installation

python3 -m pip install -r faceApi/requirments.txt

test faceAPI

edit the video path inside the test_face_Api.py file and run it: python3 test_face_Api.py