基于深度学习的复杂图案印花织物疵点检测

Complex Pattern Printed Fabric Defect Detection Based on Deep Learning

针对复杂图案印花织物疵点检测准确率不足以及误检率和漏检率较高的问题,提出了一种基于深度学习的复杂图案印花织物疵点检测方法。基于ResNet50网络融入了可变形卷积结构,获得ResNet50⁃DCN网络模型,提高了模型对不规则织物疵点的适应能力。本研究使用BFP⁃FPN模块,解决了FPN网络中存在的不同层级特征不平衡的问题,并且在BFP⁃FPN模块的输出部分引入特征融合模块,有效地消除了复杂图案花色背景的影响,并结合优化过的损失函数保证了数据处理的正确性和稳定性。本研究改进算法的mAP值67.87%,准确率93.56%,误检率与漏检率分别为6.44%、1.38%。认为:改进的复杂图案印花织物疵点检测模型有效提高了mAP值和准确率,降低了误检率和漏检率。

Aiming at the problems of insufficient detection accuracy and higher false detection rate and missed detection rate of complex pattern printed fabric defect detection,a complex pattern printed fabric defect detection method based on deep learning was proposed.Based on the ResNet50 network integrated into the DCN structure,the ResNet50-DCN network model was obtained,which could improve the model′s ability to adapt to different shape defects.The BFP-FPN module was used to solve the problem of imbalance of different levels of features in the FPN network,feature fusion module was added behind the BFP⁃FPN module,which effectively eliminated the influence of complex patterns backgrounds,and combined with the optimized loss function to ensure the correctness and stability of data processing.The mAP value of the improved algorithm in this paper was 67.87%,accuracy rate was 93.56%,error rate and missed rate were 6.44%and 1.38%.It is considered that the improved complex pattern printed fabric defect detection model could effectively improve the mAP value and accuracy,the false detection rate and the missed detection rate could be reduced.