Realtime Weapon Detection and API (Jun '20 - Jul '20)
- Constructed and trained an end-to-end YOLOV3(Darknet) model using tensorflow. Detected weapons in an image with a loss of 0.56
- Created an API for the model using Python Flask, containerized it using docker and deloyed it on Kubernetes
- Performed realtime detection using webcam in GPU and obtained a FPS of 21.3
Future Work
- Implement pose detection using OpenPose for suspicious person detection in surveillance camera.