Fast Opium Poppy Detection in Unmanned Aerial Vehicle (UAV) Imagery Based on Deep Neural Network

Opium poppy is a medicinal plant, and its cultivation illegal without legal approval in China. Unmanned aerial vehicle (UAV) an effective tool for monitoring cultivation. However, targets often appear occluded confused, it difficult existing detectors to accurately detect poppies. To address this problem, we propose opium detection network, YOLOHLA, UAV remote sensing images. Specifically, new attention module that uses two branches extract features at different scales. enhance generalization capabilities, introduce learning strategy involves iterative learning, where challenging samples are identified the model’s representation capacity enhanced using prior knowledge. Furthermore, lightweight model (YOLOHLA-tiny) YOLOHLA based on structured pruning, which can be better deployed low-power embedded platforms. evaluate performance of proposed method, collect image dataset. The experimental results show achieves faster execution speed than models. Our method mean average precision (mAP) 88.2% F1 score 85.5% detection. inference 172 frames per second (FPS) showcase practical applicability object real-time