🎯 People Detection & Tracking using YOLOv8 + ByteTrack

This repository showcases an experimental pipeline I explored to understand the integration of YOLOv8 object detection and ByteTrack multi-object tracking. It was a foundational step before developing a more robust and production-ready system.

---

🚀 Overview

This project focuses on detecting and tracking people in videos using:

• YOLOv8 (Ultralytics) — for real-time object detection
• ByteTrack — for associating detections across video frames to generate unique identities

---

📋 Project Highlights

• ✅ Detects only the person class from videos
• ✅ Draws bounding boxes with class names and confidence scores
• ✅ Saves detections in a ByteTrack-compatible format
• ✅ Tracks people across frames with unique IDs
• ✅ Exports annotated videos and tracking logs

---

📂 Outputs

• output_people_detected.mp4 — video with YOLOv8 detection only
• output_tracked.mp4 — video with YOLO + ByteTrack tracking
• bytetrack_results.txt — tracking results in MOT format
• yolo_dets/video.txt — raw detection data for tracking input

---

🔧 Environment

• Developed and tested in Google Colab
• Uses open-source models and tracking algorithms
• Includes setup for handling compatibility issues with NumPy and dependencies

---

📌 Note

This was an early-stage exploratory attempt to prototype people tracking from videos. The goal was to better understand the workings of object detection and tracking frameworks before moving on to a more advanced implementation.

---

🤝 Acknowledgements

• Ultralytics — YOLOv8
• ByteTrack — Multi-object tracking