Digital Fencing for Farms: Enhancing Object Detection through Multi-Dataset Integration

Security in farm areas is a critical concern due to the risks posed by wild animal invasions, farm animals straying, and unauthorized human entry. Conventional object detection models, often trained on urban or context-specific datasets, frequently underperform in farm settings. To address this, our study investigates the integration of multi-datasets to enhance object detection for digital fences, thereby improving farm security. We propose a method to effectively utilize multi-datasets, even when they do not have the same classes, ensuring comprehensive coverage of all required categories. The proposed SmartClass methodology achieved more robust and adaptable detection approaches, suitable for agricultural environments, with considerable increases in recall, mAP50, and mAP50-95 metrics compared to models trained without the methodology. The code and data are available at github.com/MaVILab-UFV/Digital-Fencing-WVC-2024.