Effective Pest Detection by YOLOv5 with BottleneckCSP

Pest damage poses a significant risk to crop quality and productivity, underscoring the necessity for accurate and efficient detection methods for farmers. Artificial intelligence (AI) facilitates targeted pest management with reduced reliance on broad-spectrum insecticides and enables early detection of insect infestations. Object detection, a computer vision approach, is revolutionizing pest identification in agriculture by locating pests in digital crop photographs. However, the detection of tiny objects within these datasets presents challenges due to sparse spatial information. In response to this challenge, we have replaced the C3 block of YOLOv5 with the BottleneckCSP module, which is more effective for feature mapping of tiny objects. Our enhanced model demonstrates superior performance on the IP102 dataset, which is characterized by the presence of tiny objects. Comparative analysis reveals that our BottleneckCSP-integrated model significantly improves feature mapping, resulting in a notable increase in mean average precision (mAP). Specifically, our modified model achieves an mAP of 63.9%, outperforming many state-of-the-art methods on this dataset. This advancement in the YOLOv5 architecture sets a new benchmark for AI-powered precision agriculture, providing farmers with a highly efficient tool for crop protection and pest management. The significant performance gains highlight the potential of our approach to transform current agricultural practices, paving the way for more sustainable and effective pest control strategies.