Fabric defect detection with an attention mechanism based on hard sample training

In view of the various types of fabric defects, and the problems of confusion, density unevenness and small target defects, which are difficult to detect, this paper builds a deep learning defect detection network incorporating an attention mechanism. The data augmentation strategy is used to enrich the number of samples of each defective type, and the enriched samples were extracted by the feature extraction network integrated with the attention mechanism, which can improve the feature extraction ability of confusable defect types and small defect types. Region proposal generation generates a proposal box for extracted features, and adds an online hard example mining strategy to re-learn hard examples to accelerate network convergence. Region feature aggregation maps the proposal box to the feature map to obtain the region of interest. Finally, the defect features are classified and the bounding boxes are regressed. The results show that this algorithm can effectively detect 39 categories of fabric defects with a detection speed of 0.085 s and a detection accuracy of 0.9338.

Download Full-text

Visual Detection of Surface Defects Based on Self-Feature Comparison in Robot 3-D Printing

Applied Sciences ◽

10.3390/app10010235 ◽

2019 ◽

Vol 10 (1) ◽

pp. 235 ◽

Cited By ~ 2

Author(s):

Hongyao Shen ◽

Wangzhe Du ◽

Weijun Sun ◽

Yuetong Xu ◽

Jianzhong Fu

Keyword(s):

Feature Extraction ◽

3D Printing ◽

Defect Detection ◽

Point Cloud ◽

Fused Deposition Modeling ◽

Surface Defects ◽

Detection Accuracy ◽

Feature Extraction Method ◽

Fused Deposition ◽

Detection Technology

Fused Deposition Modeling (FDM) additive manufacturing technology is widely applied in recent years. However, there are many defects that may affect the surface quality, accuracy, or even cause the collapse of the parts in the printing process. In the existing defect detection technology, the characteristics of parts themselves may be misjudged as defects. This paper presents a solution to the problem of distinguishing the defects and their own characteristics in robot 3-D printing. A self-feature extraction method of shape defect detection of 3D printing products is introduced. Discrete point cloud after model slicing is used both for path planning in 3D printing and self-feature extraction at the same time. In 3-D printing, it can generate G-code and control the shooting direction of the camera. Once the current coordinates have been received, the self-feature extraction begins, whose key steps are keeping a visual point cloud of the printed part and projecting the feature points to the picture under the equal mapping condition. After image processing technology, the contours of pictured projected and picture captured will be detected. At last, the final defects can be identified after evaluation of contour similarity based on empirical formula. This work will help to detect the defects online, improve the detection accuracy, and reduce the false detection rate without being affected by its own characteristics.

Download Full-text

A Convolutional Neural Network Based on Grouping Structure for Scene Classification

Remote Sensing ◽

10.3390/rs13132457 ◽

2021 ◽

Vol 13 (13) ◽

pp. 2457

Author(s):

Xuan Wu ◽

Zhijie Zhang ◽

Wanchang Zhang ◽

Yaning Yi ◽

Chuanrong Zhang ◽

...

Keyword(s):

Neural Network ◽

Remote Sensing ◽

Feature Extraction ◽

Convolutional Neural Network ◽

Data Augmentation ◽

Remote Sensing Image ◽

Difficult Problem ◽

Image Features ◽

Attention Mechanism ◽

Scene Classification

Convolutional neural network (CNN) is capable of automatically extracting image features and has been widely used in remote sensing image classifications. Feature extraction is an important and difficult problem in current research. In this paper, data augmentation for avoiding over fitting was attempted to enrich features of samples to improve the performance of a newly proposed convolutional neural network with UC-Merced and RSI-CB datasets for remotely sensed scene classifications. A multiple grouped convolutional neural network (MGCNN) for self-learning that is capable of promoting the efficiency of CNN was proposed, and the method of grouping multiple convolutional layers capable of being applied elsewhere as a plug-in model was developed. Meanwhile, a hyper-parameter C in MGCNN is introduced to probe into the influence of different grouping strategies for feature extraction. Experiments on the two selected datasets, the RSI-CB dataset and UC-Merced dataset, were carried out to verify the effectiveness of this newly proposed convolutional neural network, the accuracy obtained by MGCNN was 2% higher than the ResNet-50. An algorithm of attention mechanism was thus adopted and incorporated into grouping processes and a multiple grouped attention convolutional neural network (MGCNN-A) was therefore constructed to enhance the generalization capability of MGCNN. The additional experiments indicate that the incorporation of the attention mechanism to MGCNN slightly improved the accuracy of scene classification, but the robustness of the proposed network was enhanced considerably in remote sensing image classifications.

Download Full-text

An Improved YOLOv2 for Vehicle Detection

Sensors ◽

10.3390/s18124272 ◽

2018 ◽

Vol 18 (12) ◽

pp. 4272 ◽

Cited By ~ 28

Author(s):

Jun Sang ◽

Zhongyuan Wu ◽

Pei Guo ◽

Haibo Hu ◽

Hong Xiang ◽

...

Keyword(s):

Feature Extraction ◽

Vehicle Detection ◽

Training Dataset ◽

Validation Dataset ◽

Detection Accuracy ◽

Extraction Ability ◽

Detection Model ◽

Vehicle Type ◽

Proposed Model ◽

Bounding Boxes

Vehicle detection is one of the important applications of object detection in intelligent transportation systems. It aims to extract specific vehicle-type information from pictures or videos containing vehicles. To solve the problems of existing vehicle detection, such as the lack of vehicle-type recognition, low detection accuracy, and slow speed, a new vehicle detection model YOLOv2_Vehicle based on YOLOv2 is proposed in this paper. The k-means++ clustering algorithm was used to cluster the vehicle bounding boxes on the training dataset, and six anchor boxes with different sizes were selected. Considering that the different scales of the vehicles may influence the vehicle detection model, normalization was applied to improve the loss calculation method for length and width of bounding boxes. To improve the feature extraction ability of the network, the multi-layer feature fusion strategy was adopted, and the repeated convolution layers in high layers were removed. The experimental results on the Beijing Institute of Technology (BIT)-Vehicle validation dataset demonstrated that the mean Average Precision (mAP) could reach 94.78%. The proposed model also showed excellent generalization ability on the CompCars test dataset, where the “vehicle face” is quite different from the training dataset. With the comparison experiments, it was proven that the proposed method is effective for vehicle detection. In addition, with network visualization, the proposed model showed excellent feature extraction ability.

Download Full-text

Solder Joint Defect Detection in the Connectors Using Improved Faster-RCNN Algorithm

Applied Sciences ◽

10.3390/app11020576 ◽

2021 ◽

Vol 11 (2) ◽

pp. 576

Author(s):

Kaihua Zhang ◽

Haikuo Shen

Keyword(s):

Solder Joint ◽

Defect Detection ◽

Data Augmentation ◽

Detection Method ◽

Detection Algorithm ◽

Detection Methods ◽

Detection Accuracy ◽

Data Set ◽

Electronic Products ◽

Original Algorithm

The miniaturization and high integration of electronic products have higher and higher requirements for welding of internal components of electronic products. A welding quality detection method has always been one of the important research contents in the industry, among which, the research on solder joint defect detection of a connector has gradually attracted people’s attention with the development of image detection algorithm. The traditional solder joint detection method of connector adopts manual detection or automatic detection methods, which is inefficient and not safe enough. With the development of deep learning, the application of a deep convolutional neural network to target detection has become a research hotspot. In this paper, a data set of connector solder joint samples was made and the number of image samples was expanded to more than 3 times of the original by using data augmentation. Clustering generates anchor boxes and transfer learning with ResNet-101 were fused, so an improved faster region-based convolutional neural networks (Faster RCNN) algorithm was proposed. The experiment verified that the improved algorithm proposed in this paper had a great improvement in all aspects compared with the original algorithm. The average detection accuracy of this method can reach 94%, and the detection rate of some defects can even reach 100%, which can completely meet the industrial requirements.

Download Full-text

Crystallographic Symmetry for Data Augmentation in Detecting Dendrite Cores

Electronic Imaging ◽

10.2352/issn.2470-1173.2020.14.coimg-248 ◽

2020 ◽

Vol 2020 (14) ◽

pp. 248-1-248-7

Author(s):

Lan Fu ◽

Hongkai Yu ◽

Megna Shah ◽

Jeff Simmons ◽

Song Wang

Keyword(s):

Cross Sections ◽

Large Scale ◽

Data Augmentation ◽

Ground Truth ◽

Training Data ◽

Control Procedure ◽

Detection Accuracy ◽

Crystallographic Symmetry ◽

Augmentation Strategy ◽

Series Of Experiments

Accurately and rapidly detecting the locations of the cores of large-scale dendrites from 2D sectioned microscopic images helps quantify the microstructure of material components. This provides a critical link between the processing and properties of the material. Such a tool could be a critical part of a quality control procedure for manufacturing these components. In this paper, we propose to use Faster R-CNN, a convolutional neural network (CNN) model that considers both the detection accuracy and computational efficiency, to detect the dendrite cores with complex shapes. However, training CNN models usually requires a large number of images annotated with ground-truth locations of dendrite cores, which are usually obtained by highly laborintensive manual annotations. In this paper, we leverage the crystallographic symmetry of dendrite cores for data augmentation – the cross sections of dendrite cores show, not perfect, but near four-fold rotation symmetry and we can rotate the image around the center of dendrite cores by specified angles to construct new training data without additional manual annotations. We conduct a series of experiments and the results show the effectiveness of the Faster R-CNN method with the proposed data augmentation strategy. Particularly, we find that we can reduce the number of the manually annotated training images by 75% while still maintaining the same detection accuracy of dendrite cores.

Download Full-text

Neural Network-Oriented Big Data Model for Yoga Movement Recognition

Computational Intelligence and Neuroscience ◽

10.1155/2021/4334024 ◽

2021 ◽

Vol 2021 ◽

pp. 1-10

Author(s):

Hui Wang

Keyword(s):

Neural Network ◽

Feature Extraction ◽

Data Model ◽

Region Of Interest ◽

Bilinear Interpolation ◽

Network Efficiency ◽

Convolutional Network ◽

Movement Recognition ◽

Bounding Boxes ◽

Candidate Regions

The use of computer vision for target detection and recognition has been an interesting and challenging area of research for the past three decades. Professional athletes and sports enthusiasts in general can be trained with appropriate systems for corrective training and assistive training. Such a need has motivated researchers to combine artificial intelligence with the field of sports to conduct research. In this paper, we propose a Mask Region-Convolutional Neural Network (MR-CNN)- based method for yoga movement recognition based on the image task of yoga movement recognition. The improved MR-CNN model is based on the framework and structure of the region-convolutional network, which proposes a certain number of candidate regions for the image by feature extraction and classifies them, then outputs these regions as detected bounding boxes, and does mask prediction for the candidate regions using segmentation branches. The improved MR-CNN model uses an improved deep residual network as the backbone network for feature extraction, bilinear interpolation of the extracted candidate regions using Region of Interest (RoI) Align, followed by target classification and detection, and segmentation of the image using the segmentation branch. The model improves the convolution part in the segmentation branch by replacing the original standard convolution with a depth-separable convolution to improve the network efficiency. Experimentally constructed polygon-labeled datasets are simulated using the algorithm. The deepening of the network and the use of depth-separable network improve the accuracy of detection while maintaining the reliability of the network and validate the effectiveness of the improved MR-CNN.

Download Full-text

SAAE-DNN: Deep Learning Method on Intrusion Detection

Symmetry ◽

10.3390/sym12101695 ◽

2020 ◽

Vol 12 (10) ◽

pp. 1695

Author(s):

Chaofei Tang ◽

Nurbol Luktarhan ◽

Yuxin Zhao

Keyword(s):

Machine Learning ◽

Feature Extraction ◽

Random Forest ◽

Intrusion Detection ◽

Decision Tree ◽

Binary Classification ◽

Attention Mechanism ◽

Detection Methods ◽

Detection Accuracy ◽

Multi Classification

Intrusion detection system (IDS) plays a significant role in preventing network attacks and plays a vital role in the field of national security. At present, the existing intrusion detection methods are generally based on traditional machine learning models, such as random forest and decision tree, but they rely heavily on artificial feature extraction and have relatively low accuracy. To solve the problems of feature extraction and low detection accuracy in intrusion detection, an intrusion detection model SAAE-DNN, based on stacked autoencoder (SAE), attention mechanism and deep neural network (DNN), is proposed. The SAE represents data with a latent layer, and the attention mechanism enables the network to obtain the key features of intrusion detection. The trained SAAE encoder can not only automatically extract features, but also initialize the weights of DNN potential layers to improve the detection accuracy of DNN. We evaluate the performance of SAAE-DNN in binary-classification and multi-classification on an NSL-KDD dataset. The SAAE-DNN model can detect normally and attack symmetrically, with an accuracy of 87.74% and 82.14% (binary-classification and multi-classification), which is higher than that of machine learning methods such as random forest and decision tree. The experimental results show that the model has a better performance than other comparison methods.

Download Full-text

Underwater Biological Detection Algorithm Based on Improved Faster-RCNN

Water ◽

10.3390/w13172420 ◽

2021 ◽

Vol 13 (17) ◽

pp. 2420

Author(s):

Pengfei Shi ◽

Xiwang Xu ◽

Jianjun Ni ◽

Yuanxue Xin ◽

Weisheng Huang ◽

...

Keyword(s):

Feature Extraction ◽

Feature Fusion ◽

Detection Algorithm ◽

Training Data ◽

Ecological Environment ◽

Detection Accuracy ◽

Biological Detection ◽

Multi Scale ◽

Feature Pyramid ◽

Bounding Boxes

Underwater organisms are an important part of the underwater ecological environment. More and more attention has been paid to the perception of underwater ecological environment by intelligent means, such as machine vision. However, many objective reasons affect the accuracy of underwater biological detection, such as the low-quality image, different sizes or shapes, and overlapping or occlusion of underwater organisms. Therefore, this paper proposes an underwater biological detection algorithm based on improved Faster-RCNN. Firstly, the ResNet is used as the backbone feature extraction network of Faster-RCNN. Then, BiFPN (Bidirectional Feature Pyramid Network) is used to build a ResNet–BiFPN structure which can improve the capability of feature extraction and multi-scale feature fusion. Additionally, EIoU (Effective IoU) is used to replace IoU to reduce the proportion of redundant bounding boxes in the training data. Moreover, K-means++ clustering is used to generate more suitable anchor boxes to improve detection accuracy. Finally, the experimental results show that the detection accuracy of underwater biological detection algorithm based on improved Faster-RCNN on URPC2018 dataset is improved to 88.94%, which is 8.26% higher than Faster-RCNN. The results fully prove the effectiveness of the proposed algorithm.

Download Full-text

End Image Defect Detection of Float Glass Based on Faster Region-Based Convolutional Neural Network

Traitement du signal ◽

10.18280/ts.370513 ◽

2020 ◽

Vol 37 (5) ◽

pp. 807-813

Author(s):

Dabing Jin ◽

Shiqing Xu ◽

Lianjie Tong ◽

Linyu Wu ◽

Shimin Liu

Keyword(s):

Neural Network ◽

Feature Extraction ◽

Convolutional Neural Network ◽

Production Process ◽

Defect Detection ◽

Raw Materials ◽

Float Glass ◽

Detection Accuracy ◽

Detection Model ◽

Image Defect

The float glass contains various defects for reasons of raw materials and production process. These defects can be observed on the end images of the glass. Since the defects are correlated with specific links of the production process, it is possible to discover the process problems by identifying the location and type of defects in end images. Based on faster region-based convolutional neural network (Faster RCNN), this paper proposes a deep learning method that improves the feature extraction network, and adds a Laplacian convolutional layer to preprocess the end images. Considering the defect features in end images, the anchor box size was adjusted to speed up the training. Besides, the lack of generalizability induced by small dataset was solved through data enhancement. With improved VGG16 as the feature extraction layer, a glass defect detection model was established, whose generalizability was improved through transfer learning. The experimental results show that the proposed model achieved a mean detection accuracy of 94% on actual test set, meeting the requirements for actual use in factories.

Download Full-text

The Traffic Sign Detection Algorithm Based on Region of Interest Extraction and Double Filter

Recent Advances in Computer Science and Communications ◽

10.2174/2213275912666190823112357 ◽

2019 ◽

Vol 13 ◽

Author(s):

Dongxian Yu ◽

Jiatao Kang ◽

Zaihui Cao ◽

Neha Jain

Keyword(s):

Region Of Interest ◽

Detection Algorithm ◽

Light Filter ◽

Support Vector ◽

Detection Accuracy ◽

Traffic Light ◽

Traffic Sign ◽

Traffic Signs ◽

Sign Detection ◽

Traffic Sign Detection

In order to solve the current traffic sign detection technology due to the interference of various complex factors, it is difficult to effectively carry out the correct detection of traffic signs, and the robustness is weak, a traffic sign detection algorithm based on the region of interest extraction and double filter is designed.First, in order to reduce environmental interference, the input image is preprocessed to enhance the main color of each logo.Secondly, in order to improve the extraction ability Of Regions Of Interest, a Region Of Interest (ROI) detector based on Maximally Stable Extremal Regions (MSER) and Wave Equation (WE) was defined, and candidate Regions were selected through the ROI detector.Then, an effective HOG (Histogram of Oriented Gradient) descriptor is introduced as the detection feature of traffic signs, and SVM (Support Vector Machine) is used to classify them into traffic signs or background.Finally, the context-aware filter and the traffic light filter are used to further identify the false traffic signs and improve the detection accuracy.In the GTSDB database, three kinds of traffic signs, which are indicative, prohibited and dangerous, are tested, and the results show that the proposed algorithm has higher detection accuracy and robustness compared with the current traffic sign recognition technology.

Download Full-text