This repository implements a complete Animal Re-Identification (ReID) system for Tiger and Deer, integrating:

• YOLOv11 (Ultralytics) — Fast object detection
• MegaDetector v5a.0.0 — Wildlife-focused detection
• Deep CNN feature extraction (ResNet)
• Pattern feature extraction (LBP + Color Histogram)
• Combined feature embedding
• FAISS similarity search
• SVM classifier for ReID
• Incremental learning (auto-add new animals)
• Top-1 best match retrieval with visualization

The system automatically detects animals, extracts deep + pattern features, stores them in a FAISS index, and retrieves the best match in real-time.

This project supports dual detector mode:

• Specialized for wildlife detection
• More robust in forest/camera-trap environments
• Detects broad categories: "animal", "human", "vehicle"

• Very fast
• Detects more specific species (tiger/deer) from training data
• Used when MegaDetector is unavailable or disabled

The pipeline attempts:

1. Load MegaDetector → use if available
2. Else fallback to YOLOv11

Both models are fully integrated into the detection process.