Forestpest-YOLO: A High-Performance Detection Framework for Small Forestry Pests
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866916982367977472 |
|---|---|
| author | Li, Aoduo Lin, Peikai Li, Jiancheng Zhang, Zhen Wu, Shiting Liang, Zexiao Jiang, Zhifa |
| author_facet | Li, Aoduo Lin, Peikai Li, Jiancheng Zhang, Zhen Wu, Shiting Liang, Zexiao Jiang, Zhifa |
| contents | Detecting agricultural pests in complex forestry environments using remote sensing imagery is fundamental for ecological preservation, yet it is severely hampered by practical challenges. Targets are often minuscule, heavily occluded, and visually similar to the cluttered background, causing conventional object detection models to falter due to the loss of fine-grained features and an inability to handle extreme data imbalance. To overcome these obstacles, this paper introduces Forestpest-YOLO, a detection framework meticulously optimized for the nuances of forestry remote sensing. Building upon the YOLOv8 architecture, our framework introduces a synergistic trio of innovations. We first integrate a lossless downsampling module, SPD-Conv, to ensure that critical high-resolution details of small targets are preserved throughout the network. This is complemented by a novel cross-stage feature fusion block, CSPOK, which dynamically enhances multi-scale feature representation while suppressing background noise. Finally, we employ VarifocalLoss to refine the training objective, compelling the model to focus on high-quality and hard-to-classify samples. Extensive experiments on our challenging, self-constructed ForestPest dataset demonstrate that Forestpest-YOLO achieves state-of-the-art performance, showing marked improvements in detecting small, occluded pests and significantly outperforming established baseline models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_00547 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Forestpest-YOLO: A High-Performance Detection Framework for Small Forestry Pests Li, Aoduo Lin, Peikai Li, Jiancheng Zhang, Zhen Wu, Shiting Liang, Zexiao Jiang, Zhifa Computer Vision and Pattern Recognition Artificial Intelligence Detecting agricultural pests in complex forestry environments using remote sensing imagery is fundamental for ecological preservation, yet it is severely hampered by practical challenges. Targets are often minuscule, heavily occluded, and visually similar to the cluttered background, causing conventional object detection models to falter due to the loss of fine-grained features and an inability to handle extreme data imbalance. To overcome these obstacles, this paper introduces Forestpest-YOLO, a detection framework meticulously optimized for the nuances of forestry remote sensing. Building upon the YOLOv8 architecture, our framework introduces a synergistic trio of innovations. We first integrate a lossless downsampling module, SPD-Conv, to ensure that critical high-resolution details of small targets are preserved throughout the network. This is complemented by a novel cross-stage feature fusion block, CSPOK, which dynamically enhances multi-scale feature representation while suppressing background noise. Finally, we employ VarifocalLoss to refine the training objective, compelling the model to focus on high-quality and hard-to-classify samples. Extensive experiments on our challenging, self-constructed ForestPest dataset demonstrate that Forestpest-YOLO achieves state-of-the-art performance, showing marked improvements in detecting small, occluded pests and significantly outperforming established baseline models. |
| title | Forestpest-YOLO: A High-Performance Detection Framework for Small Forestry Pests |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2510.00547 |