Scholarlink

Aiming at the problem of decreasing accuracy of silkworm disease detection in the occluded environment of silkworms, this paper proposes an improved RT-DETR-SC model. Based on the original RT-DETR, the model introduces Conditionally Gated Linear Unit (CGLU) and Multiscale Attention Fusion Module (SMAFB), which are used to enhance the responsiveness to the occluded small targets and improve the feature focusing and noise suppression effects, respectively. The model performance is verified through multiple sets of ablation experiments, and the results show that RT-DETR-SC achieves 98.4% on mAP@0.5, which is 2.2 percentage points higher than the original model, and also performs well in Precision and Recall, demonstrating its recognition accuracy and robustness in complex occlusion scenes.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Detection of Sericultural Diseases in Silkworm Shading Environment Based on RT-DETR-SC

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Zilin Huang,
  • Peng Chen,
  • Tin Tin Ting,
  • Maidin Siti Sarah,
  • Xiaoyu Meng,
  • Shuaiming Qiu

摘要

Aiming at the problem of decreasing accuracy of silkworm disease detection in the occluded environment of silkworms, this paper proposes an improved RT-DETR-SC model. Based on the original RT-DETR, the model introduces Conditionally Gated Linear Unit (CGLU) and Multiscale Attention Fusion Module (SMAFB), which are used to enhance the responsiveness to the occluded small targets and improve the feature focusing and noise suppression effects, respectively. The model performance is verified through multiple sets of ablation experiments, and the results show that RT-DETR-SC achieves 98.4% on mAP@0.5, which is 2.2 percentage points higher than the original model, and also performs well in Precision and Recall, demonstrating its recognition accuracy and robustness in complex occlusion scenes.