Built multi-object tracking system for parcels and people with persistent IDs. This project demonstrates practical execution from architecture and implementation to measurable delivery outcomes.
Objective
Built multi-object tracking system for parcels and people with persistent IDs.
Stack
Delivery highlights
Demo Video use YOLO V8-seg and BoT-SORT
Watch on sourceDemo Video use YOLO V26-seg and BoT-SORT
Watch on sourceExtended previous Electric Vehicle Charger Socket Instance Segmentation and Electric Vehicle Charger Socket Instance Segmentation & Tracking projects by building a custom DAVIS-style dataset with YOLOv8-Seg and adapting it for XMem to enable temporally consistent pixel-level segmentation across video frames.
Built parcel/person segmentation benchmark with unified dataset and fair model comparison.
Built upon the previous Semi-Supervised Video Object Segmentation for Electric Vehicle Charger Socket Tracking project by applying small-object-focused data augmentation to address missed detections when charger sockets appeared smaller or farther from the camera. This improved the model’s ability to detect small and hard-to-detect sockets while reducing missed detections in real-world scenarios.