Skip to content

Latest commit

 

History

History
32 lines (21 loc) · 994 Bytes

README.md

File metadata and controls

32 lines (21 loc) · 994 Bytes
Raw

Car-Counting

To run

python main.py path-to-video

This project is pretty simple thanks to yolov8 with new implementation of tracker. You can change tracking algorithm with

  • BoT-SORT - botsort.yaml

  • ByteTrack - bytetrack.yaml

results = model.track(source="https://youtu.be/Zgi9g1ksQHc", tracker="bytetrack.yaml")

If you have GPU,

How it works

  1. Set trigger position
  2. Load the model (of course)
  3. Iterate frame by frame of the video
    1. Set unique ID for each object
    2. Check if any object cross the trigger
    3. Draw result
  4. End

YOUTUBE VIDEO

Note: this project uses supervision to detect when objects cross the trigger. It's on beta and many thing has changed so their documentation is outdated at the moment. You should watch their latest video instead.

The video used to test is recorded from camstreamer