Skip to content

Latest commit

 

History

History
14 lines (11 loc) · 675 Bytes

File metadata and controls

14 lines (11 loc) · 675 Bytes

Realtime Gender Recognition from Face

The keras model is created by training from scratch on around 2200 face images (~1100 for each class). Face region is cropped by applying face detection using cvlib on the images gathered from Kaggle. It acheived around 96% training accuracy and ~90% validation accuracy. (20% of the dataset is used for validation)

Python packages

  • numpy
  • opencv-python
  • tensorflow
  • keras
  • requests
  • progressbar
  • cvlib

We would first need to generate model by executing the train_model.ipynb file. Once the model is generated successfully,it is saved and then we can use detect_gender.ipynb to get realtime gender classification.