All systems
On GitHubMALULEKE-KS
Machine Learning Project
Real-Time Multi-Object Detection System powered by YOLOv8 - Advanced computer vision solution for identifying and tracking 80+ object classes across images, video streams, and live camera feeds with high accuracy and performance optimization.
- Python
- YOLOv8
- OpenCV
- PyTorch
- Started
- March 2026
- Last push
- 7 months ago
- Commits this year
- 4
Written by AI from the repository · updated 5 days agoAI
The problem
Recognising and following everyday objects — people, vehicles, animals — in pictures, recorded video and a live camera, with one program.
How it works
Built on YOLOv8 (Ultralytics, the yolov8n model trained on the 80 COCO classes) and OpenCV, with three modes:
- Images — detects every object, draws labelled boxes, counts them and saves the annotated image.
- Video — tracks objects frame by frame with persistent IDs, so the same car keeps the same label across frames, and writes an annotated video.
- Webcam — live detection with on-screen frames per second, an object count and each detection's confidence, recorded to a file.
Tuning
Confidence and overlap thresholds are set per mode — lower on the live camera, where objects move and blur, with the frame rate capped for steadier tracking.
More systems
