We should be able to identify subjects from CCTV footages even when there are only limited number of photographs available for the suspected individuals. The system should work in near real time and even when there are partial occlusion, different pose, and physical changes like beard , eye-glasses and face masks .
To deploy this project there are 2 options :
- Run on COLAB : Open Colab folder in the repo
- Run Locally : Coming Soon
Install these libraries
- Recognition at varying angles
- Recognition of people wearing surgical masks due to covid-19
- Identifying using limited suspect images
- No training requied to be done on dataset (dynamic recognition)
Contributions are always welcome!