Published Paper Details:

Recent advances in deep learning--particularly Convolutional Neural Networks (CNNs)--have opened transformative opportunities for automated object detection and classification in complex outdoor environments. Models from the YOLO (You Only Look Once) family, especially YOLOv8, have emerged as highly effective tools for real-time multi-class detection, offering an optimal balance between inference speed and detection accuracy. Unlike rule-based or template-matching methods, CNN-based architectures learn intricate visual features directly from data, enabling them to adapt to varying lighting conditions, occlusion from vegetation, and complex natural backgrounds [2]. This paper proposes a complete smart farm surveillance framework that combines YOLOv8-based object detection with automated Telegram Bot API notifications. The system processes live video feeds from farm cameras, identifies threats as they occur, captures annotated snapshots, and delivers instant alerts to farmers' smartphones. Key contributions include: o A fine-tuned YOLOv8 detection model trained on a diverse agricultural dataset covering humans, birds, and animals. o An integrated Telegram Bot alerting module providing real-time notifications with timestamped annotated images. o Deployment on resource-constrained edge devices (Raspberry Pi / NVIDIA Jetson Nano) demonstrating practical feasibility. o Comprehensive experimental evaluation under varied farm conditions confirming 94.7% precision and 93.1% recall.