face_classification

Real-time face detection and emotion/gender classification

Input

Ailia input shape: (1, 1, 64, 64)
Range: [-1.0, 1.0]

Output

emotion_class_count=3
+ idx=0
  category=6[ neutral ]
  prob=0.411855548620224
+ idx=1
  category=4[ sad ]
  prob=0.1994263231754303
+ idx=2
  category=0[ angry ]
  prob=0.19452838599681854

gender_class_count=2
+ idx=0
  category=0[ female ]
  prob=0.8007728457450867
+ idx=1
  category=1[ male ]
  prob=0.19922710955142975

Usage

Automatically downloads the onnx and prototxt files on the first run. It is necessary to be connected to the Internet while downloading.

For the sample image,

$ python3 face_classification.py 

If you want to specify the input image, put the image path after the --input option.

$ python3 face_classification.py --input IMAGE_PATH

By adding the --video option, you can input the video.
If you pass 0 as an argument to VIDEO_PATH, you can use the webcam input instead of the video file.

$ python3 face_classification.py --video VIDEO_PATH

Reference

Real-time face detection and emotion/gender classification

Framework

Keras

Model Format

CaffeModel

Netron

emotion_miniXception.prototxt