In the face of detection problems posed by complex textile texture backgrounds, different sizes, and different types of defects, commonly used object detection networks have limitations in handling target sizes. Furthermore, their stability and anti-jamming capabilities are relatively weak. Therefore, when the target types are more diverse, false detections or missed detections are likely to occur. In order to meet the stringent requirements of textile defect detection, we propose a novel AC-YOLOv5-based textile defect detection method. This method fully considers the optical properties, texture distribution, imaging properties, and detection requirements specific to textiles. First, the Atrous Spatial Pyramid Pooling (ASPP) module is introduced into the YOLOv5 backbone network, and the feature map is pooled using convolution cores with different expansion rates. Multiscale feature information is obtained from feature maps of different receptive fields, which improves the detection of defects of different sizes without changing the resolution of the input image. Secondly, a convolution squeeze-and-excitation (CSE) channel attention module is proposed, and the CSE module is introduced into the YOLOv5 backbone network. The weights of each feature channel are obtained through self-learning to further improve the defect detection and anti-jamming capability. Finally, a large number of fabric images were collected using an inspection system built on a circular knitting machine at an industrial site, and a large number of experiments were conducted using a self-built fabric defect dataset. The experimental results showed that AC-YOLOv5 can achieve an overall detection accuracy of 99.1% for fabric defect datasets, satisfying the requirements for applications in industrial areas.

Automatic vision-based defect detection on the mobile light guide plate (LGP) is a challenging task due to the low contrast between the defect and the background, uneven brightness, and complex gradient texture. An end-to-end multitask learning network architecture for the defect detection of mobile phone LGP is proposed. First, the main structure of the multitask learning network is designed. The encoder part uses a similar U-Net encoder structure to obtain multiscale features, and the feature fusion part adopts feature fusion to interact with multiscale features. Second, the segmentation head is designed to complete the precise location of each defect in an image by using the multiscale feature fusion, which prepares it for the quantification of defect characteristics. Combining the multiscale features and the output mask of the segmentation head, the classification head is designed to accurately detect the defects of mobile phone LGP. Finally, the defect detection data set has been constructed based on the mobile phone LGP images collected on the industrial site, and a lot of experiments are performed on the mobile phone LGP data set and Kolektor surface-defect data set (KolektorSDD). The experimental results show that the F1-score on the two data sets can reach 99.67% and 96.77%, respectively, which verifies the effectiveness of the method proposed in this article.

Aiming at the difference in density between the light guide point distribution of the light guide plate (LGP) images, the different size, shape and brightness of LGP defects, and the limitation of a small number of defect samples, our paper proposes a two-stage multiscale residual attention network based on “segmentation + decision” for LGP defect detection. In the first stage, a segmentation subnet is constructed to achieve accurate location of multiscale defects by using the U-shaped structure and the designed Multiscale Residual Attention Unit (MRAU). In the second stage, a decision subnet is constructed to achieve accurate decision whether the LGP image is defective under the guidance of features extracted by segmentation subnet. The design of the network enables the model to be trained using a small number of samples, which is a significant requirement for practical applications. Finally, a lot of experimental studies were conducted on two datasets, including compared with related deep learning methods, proving the effectiveness of residual attention mechanism and proving the effectiveness of multiscale mechanism. Experimental results show that the detection method in this paper has advantages in both detection accuracy and universality under the limitation of a small number of samples. In particular, the detection indicator F1-Score respectively reached 99.85% and 95.46% on the two datasets. Combined with an image acquisition platform, a surface defect detection system is implemented to realize the real-time production quality inspection on the production lines.

Aiming at the vehicle navigation light guide plate (LGP) image characteristics, such as complex and gradient textures, uneven brightness, and small defects, this paper proposes a visual inspection method for LGP defects based on an improved RetinaNet. First, we use ResNeXt50 with higher accuracy under the same parameters as the backbone network, and propose the lightweight module Ghost_module to replace the 1 × 1 convolution in the lower half of the ResNeXt_block. This can reduce the resource parameters and consumption, and speed up training and inference. Second, we propose and use an improved feature pyramid network module to improve the feature fusion network in RetinaNet. It can more effectively fuse the shallow semantic information and high-level semantic information in the backbone feature extraction network, and further improve the detection ability of small target defects. Finally, the defect detection dataset constructed based on the vehicle LGP images collected at a industrial site, and experiments are performed on the vehicle LGP dataset and Aluminum Profile Defect Identification dataset (Aluminum Profile DID). The experimental results show that the proposed method is both efficient and effective. It achieves a better average detection rate of 98.6% on the vehicle LGP dataset. The accuracy and real-time performance can meet the requirements of industrial detection.