/VideoAnalytics

An approach towards analysing road traffic

Primary LanguagePythonApache License 2.0Apache-2.0

Video Analytics for traffic intelligence

Supports:

  • Object Detection

  • Object Tracking

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

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

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

  3. 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