CV-advance_object_tracking

Notebook-based multi-object detection and tracking project using Faster R-CNN on the MOT20 dataset.

What the notebook covers

• Setup and MOT20 data download/preprocessing
• Faster R-CNN (ResNet-50 FPN) initialization and fine-tuning for person detection
• Temporal consistency filtering to reduce unstable detections
• SORT tracking with Kalman filter + IoU association
• Baseline vs fine-tuned evaluation with Precision, Recall, F1, mAP, MOTA, and MOTP
• Visual analysis of detections, tracks, ID switches, and comparison plots

Main file: advance_object_detection_fasterRCNN.ipynb