Real-time Package Counting and Multi-Object Tracking System Using YOLOv11 and Deep SORT 📦🎥
🚀 Real-time Package Counting and Multi-Object Tracking System Using YOLOv11 and Deep SORT 📦🎥
I built an end-to-end computer vision pipeline that performs precise package detection, classification by size, and counting in video streams. The system leverages:
🧠 YOLOv11 for efficient and accurate object detection with confidence filtering
🎯 Deep SORT for robust multi-object tracking and consistent ID assignment
📏 Dynamic size classification based on bounding box area thresholds (Small, Medium, Large)
➡️ Line-crossing logic to increment counters when tracked objects pass a predefined vertical boundary
👁️ Real-time visualization with semi-transparent overlays, bounding boxes, track IDs, and FPS display
💾 Output video recording with all annotations for offline analysis
Key technical components include Python, OpenCV for video processing, Ultralytics YOLO for detection, and Deep SORT for tracking.
This project helped deepen my expertise in deploying real-time vision systems combining detection and tracking, and designing reliable object counting mechanisms under varying video conditions.
Feel free to connect or discuss AI, computer vision, or multi-object tracking applications. 🤝