Human-wildlife conflict (HWC) causes an estimated 40% crop loss in border farming communities across South Asia, with annual damages exceeding ₹12,000 crore in India alone. Traditional deterrent methods including electric fencing, manual vigilance, and passive infrared (PIR) sensors suffer from high false-positive rates (up to 28%), lack of species specificity, and prohibitive maintenance costs. This paper presents an autonomous, solar-powered, edge-AI system that integrates YOLOv8 object detection, adaptive multi-modal repellent actuation, and real-time GSM-based remote monitoring for precision crop protection. The system employs a Raspberry Pi 4B edge processor running a quantized and pruned YOLOv8n model (12 MB, 0.93 mAP@0.5) that detects elephants, wild boars, and other crop-raiding species at 280 ms latency. Upon detection, species-specific repellent protocols are activated: infrasonic deterrence (10–20 Hz) for elephants, ultrasonic emission (18–25 kHz) for boars, and stroboscopic lights for nocturnal intrusions. Field trials conducted over a 10-acre farm in Tamil Nadu demonstrated a 92% animal repulsion rate, 99.7% system uptime, and a 4.2% false-positive rate—outperforming all baseline methods. A five-year cost-benefit analysis confirms a 340% return on investment versus traditional electric fencing. The proposed system demonstrates that edge AI can deliver sustainable, farmer-affordable, and ecologically non-lethal wildlife management at scale.

 Keywords: YOLOv8, Edge AI, Human-Wildlife Conflict, Crop Protection, Solar-Powered IoT, Precision Agriculture, Wild Animal Deterrent