nikheelpandey / VideoAnalytics

An approach towards analysing road traffic

Video Analytics for traffic intelligence

Supports:

• Object Detection

• Object Tracking

How to use the repo

1. Create dirs:

• model : For storing model.weight and model.cfg
• data : For storing the footage to be analysed
• result: For saving inferences with bounding boxes

2. Run json_maker.py for storing the inferences on a video feed in the file called detection.json

3. Run demo.py to use these detection on the video feed and store the inferences in result

4. Run create_video.py to stitch the images from result into a video file.

To do:

• Create end-to-end detection and tracking pipeline
• Integrate MongoDB as database
• Create deterministic logic for video analytics