Method and system for rapid detection and disposal of garbage overflow for inspection robot

CN119169560BActive Publication Date: 2026-08-28CHONGQING UNIV OF POSTS & TELECOMM
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202411277366.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-12
Publication Date
2026-08-28
Estimated Expiration
2044-09-12

AI Technical Summary

Technical Problem

然而，YOLO模型通常比较庞大，计算量大，不适合在资源受限的设备上运行，即，YOLO模型与巡检机器人结合执行垃圾满溢检测受到其庞大的计算量限制，阻碍了发展

Benefits of technology

[0048]本发明具有多个优势。首先，本发明的轻量化YOLOv8-garbage_bin模型采用幻影卷积GhostConv模块实现了模型的轻量化设计，在保持较高检测精度的同时，显著降低了参数量和计算成本。其次，本发明的轻量化YOLOv8-garbage_bin模型中的C2f_EMA模块能够在增加少量计算开销的情况下，显著提升模型的特征表达能力和泛化性能。最后，本发明采用Wise-IoU损失函数对轻量化YOLOv8-garbage_bin模型进行训练，有效减少了高质量锚框的竞争和低质量样本带来的有害梯度，从而提升了检测器的整体性能，进一步提高了模型的检测准确性。

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119169560B_ABST
    Figure CN119169560B_ABST
Patent Text Reader

Abstract

The present application relates to a kind of garbage overflow rapid detection processing method and system for inspection robot, belong to the technical field of image processing and target detection identification of inspection robot.It is used to execute the method, including: through inspection robot, collection road garbage bin image data, and the data is preprocessed;Lightweight model network is constructed, at least including GhostConv module and several C2f_EMA modules;Improved loss function is used as the loss function of the lightweight model network constructed;The image data after pre-processing is used to train the lightweight model network constructed, and garbage overflow rapid detection model is obtained;Garbage overflow rapid detection model is carried on inspection robot, and garbage overflow detection is carried out to the image collected by inspection robot in real time.The present application has higher precision while having more optimal lightweight performance.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Improved YOLOv5 lightweight community scene pedestrian detection method

    CN115862066A

  • Traffic scene target detection method and system based on improved YOLOv5s

    CN118314320A