| 摘 要: 在工业现场采用传统机器视觉技术难以满足高精度和高稳定性的检测要求,为应对电阻片车间复杂多变的环境影响,提出了一种结合图像预处理方法与YOLOv8-OBB深度学习算法的解决方案,用于实现产线上匣钵容器的检测定位。具体而言,开发了一种基于CLAHE融合边界检测模块 RCF的图像增强算法,以突出前背景边界信息并提升整体图像质量,同时,通过网络模型中主干网络的通道剪枝和检测头的轻量化设计,在保证检测精度的前提下,使 ONNX模型尺寸缩小至原来的57%。 |
| 关键词: 深度学习 旋转目标检测 图像增强 模型轻量化 |
|
中图分类号:
文献标识码: A
|
| 基金项目: 教育部人文社会科学研究一般项目(23YJCZH281);上海市哲学社会科学规划课题(2022ZGL010);信息网络安全公安部重点实验室开放课题 |
|
| Research on Adaptive Detection of Sagger Containers in Resistor Production Lines Based on Deep Learning |
|
REN Kecheng1,2, XIANG Zhong1,2
|
(1.School of Mechanical Engineering, Zhejiang Sc-i Tech University, Hangzhou 310018, China;
2.Zhejiang Sc-i Tech University Pingyang Research Institute Company, Wenzhou 325400, China)
1462619741@qq.com; xz@zstu.edu.cn
|
| Abstract: Traditional machine vision techniques struggle to meet high-precision and high-stability detection requirements in industrial settings. To address the complex and variable environmental impacts in resistor production workshops, this paper proposes a solution combining image preprocessing methods with the YOLOv8-OBB deep learning algorithm for detecting and locating sagger containers on production lines. Specifically, an image enhancement algorithm was developed based on CLAHE fused with the RCF boundary detection module to highlight foreground edge information and improve overall image quality. Additionally, through channel pruning in the backbone network and lightweight design of the detection head, the ONNX model size was reduced to 57% of the original while maintaining detection accuracy |
| Keywords: deep learning oriented object detection image enhancement model lightweighting |
