Search

In industrial applications, robotic arm grasp detection tasks frequently suffer from inadequate accuracy and success rates, which result in reduced operational efficiency. Although existing methods have achieved some success, limitations remain in terms of detection accuracy, real-time performance, and generalization ability. To address these challenges, this paper proposes an enhanced grasp detection model, G-RCenterNet, based on the CenterNet framework. First, a channel and spatial attention mechanism is introduced to improve the network's capability to extract target features, significantly enhancing grasp detection performance in complex backgrounds. Second, an efficient attention module search strategy is proposed to replace traditional fully connected layer structures, which not only increases detection accuracy but also reduces computational overhead. Additionally, the GSConv module is incorporated during the prediction decoding phase to accelerate inference speed while maintaining high accuracy, further improving real-time performance. Finally, ResNet50 is selected as the backbone network, and a custom loss function is designed specifically for grasp detection tasks, which significantly enhances the model's ability to predict feasible grasp boxes. The proposed G-RCenterNet algorithm is embedded into a robotic grasping system, where a structured light depth camera captures target images, and the grasp detection network predicts the optimal grasp box. Experimental results based on the Cornell Grasp Dataset and real-world scenarios demonstrate that the G-RCenterNet model performs robustly in grasp detection tasks, achieving accurate and efficient target grasp detection suitable for practical applications. [ABSTRACT FROM AUTHOR]

In order to realize real‐time detection of aluminum defects during aluminum production, the target detection algorithm needs to be able to run on locally deployed hardware. Convolutional neural networks can effectively extract representative features from high‐dimensional data such as images and videos, and capture spatial information in the data, making it easier to locate aluminum defects. Moreover, running CNN model inference on local hardware has high real‐time performance. Due to the advantages of convolutional neural network in anomaly detection, an improved CenterNet aluminum surface defect detection method was proposed. The algorithm combines common convolution and depthwise separable convolution to design a lightweight convolution module. Then, the Convolutional Block Attention Module is added to the feature extraction network to make the network better capture the rich input feature information of the image. Ultimately, the α‐DIoU loss function is implemented to enhance the precision of bounding box predictions. The experimental findings demonstrate that the proposed algorithm achieves an average detection accuracy (mAP) of 86.02%, which is 5.95% higher than the average detection accuracy of the traditional algorithm, and has a good detection effect on aluminum surface defects. Furthermore, there is an 11.9% reduction in model parameters and a 15.2% decrease in floating‐point computations, which helps to promote the deployment of embedded device platforms. © 2024 Institute of Electrical Engineers of Japan and Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

As one of the most commonly used and important data carriers, tables have the advantages of high structuring, strong readability and strong flexibility. However, in reality, tables usually present various forms, such as Excel, images, etc. Among them, the information in the table image cannot be read directly, let alone further applied. Therefore, the research related to image-based table recognition is crucial. It contains the table structure recognition and the table content recognition. Among them, table structure recognition is the most important and difficult task because the table structure is abstract and changeable. In order to address this problem, we propose an innovative table structure recognition method, named TSRDet (Table Structure Recognition based on object Detection). It includes a row-column detection method, named SACNet (StripAttention-CenterNet) and the corresponding post-processing. SACNet is an improved version of the original CenterNet. The specific improvements include the following: firstly, we introduce the Swin Transformer as the encoder to obtain the global feature map of the image. Then, we propose a plug-and-play row-column attention module, including a channel attention module and a row-column spatial attention module. It improves the detection accuracy of rows and columns by capturing long-range row-column feature maps in the image. After completing the row-column detection, this paper also designs a simple and fast post-processing to generate the table structure based on the row-column detection results. Experimental results show that for row-column detection, SACNet has high detection accuracy, even at a high IoU threshold. Specifically, when the threshold is 0.75, its mAP of row detection and column detection still exceeds 90%, which is 91.40% and 92.73% respectively. In addition, in the comparative experiment with the existing object detection methods, SACNet's performance was significantly better than that of all others. For table structure recognition, the TEDS-Struct score of TSRDet is 95.7%, which shows competitive performance in table structure recognition, and verifies the rationality and superiority of the proposed method. [ABSTRACT FROM AUTHOR]

Motion vehicle detection based on deep learning technology is currently a research hotspot in the intersection of traffic and computer science. To address challenges in dynamic vehicle detection tasks, such as multi-scale issues, overlapping targets, and the difficulty of distinguishing between dynamic and static vehicles, this paper proposes a multi-task feature fusion approach for CenterNet motion vehicle detection. Firstly, a task branch for vehicle segmentation is added to the network, forming a dualstream mechanism along with the original object detection stream. Subsequently, an appropriate method is employed to achieve feature fusion between the two streams, assisting in enhancing critical feature information in the object detection stream. Additionally, the introduction of attention mechanisms further optimizes model accuracy. On a test set created based on the UA-DETRAC public dataset, our proposed method achieves an average precision of 70%, representing a 5.8% improvement compared to the original CenterNet model. With a frame rate of 30 frames per second, our method demonstrates the best balance between speed and accuracy compared to the contrastive methods. Extensive experiments indicate that our approach performs well in motion vehicle detection tasks. [ABSTRACT FROM AUTHOR]

Rail surface defect detection is vital for railway safety. Traditional methods falter with varying defect sizes and complex backgrounds, while two-stage deep learning models, though accurate, lack real-time capabilities. To overcome these challenges, we propose an enhanced one-stage detection model based on CenterNet. We replace ResNet with ResNeXt and implement a multi-branch structure for better low-level feature extraction. Additionally, we integrate SKNet attention mechanism with the C2f structure from YOLOv8, improving the model's focus on critical image regions and enhancing the detection of minor defects. We also introduce an elliptical Gaussian kernel for size regression loss, better representing the aspect ratio of rail defects. This approach enhances detection accuracy and speeds up training. Our model achieves a mean accuracy (mAP) of 0.952 on the rail defects dataset, outperforming other models with a 6.6% improvement over the original and a 35.5% increase in training speed. These results demonstrate the efficiency and reliability of our method for rail defect detection. [ABSTRACT FROM AUTHOR]

This paper proposes a novel object detection method to address the challenges posed by small objects and occlusion in object detection. This work is performed within the CenterNet framework, leveraging the MobileNetV3 backbone to model the input image's abstract representation in a lightweight manner. A sparse convolutional skip connection is introduced in the bottleneck of MobileNetV3, specifically designed to adaptively suppress redundant and interfering information, thus enhancing feature extraction capabilities. A Dual-Path Bidirectional Feature Pyramid Network (DBi-FPN) is incorporated, allowing for high-level feature fusion through bidirectional flow and significantly improving the detection capabilities for small objects and occlusions. Task heads are applied within the feature space of multi-scale information merged by DBi-FPN, facilitating comprehensive consideration of multi-level representations. A bounding box-area loss function is also introduced, aimed at enhancing the model's adaptability to object morphologies and geometric distortions. Extensive experiments on the PASCAL VOC 2007 and MS COCO 2017 datasets validate the competitiveness of our proposed method, particularly in real-time applications on resource-constrained devices. Our contributions offer promising avenues for enhancing the accuracy and robustness of object detection systems in complex scenarios. [ABSTRACT FROM AUTHOR]

Camellia oleifera is a typical dense and small economic oil fruit. Its traditional harvesting method relies on manual labor, which is inefficient and costly. It is of significance to develop automatic harvesting machinery, which relies on rapid and accurate detection of Camellia oleifera in natural and complex environment. In this study, a novel model named CenterNet-LW-SE was proposed. Based on the structure of CenterNet, the lightweight Deep Layer Aggregation was used as the feature extraction backbone and the Squeeze-and-Excitation module was embedded to enhance the ability of feature extracting. A total of 4700 images were used to train and test the algorithms. The precision, recall, AP 75 , F 1 , model size and detection speed of CenterNet-LW-SE were 94.70%, 87.90%, 90.00%, 91.17%, 65.65 MB and 5.03 fps, respectively. Compared with the original CenterNet its AP 75 was improved by 2.1%, its model size was reduced by 16.69%, and its average detection time on the development board was reduced by 30.40%. Experimental results showed that for severely occluded, shadowy and blurred fruits, CenterNet-LW-SE had better robustness and portability than CenterNet-DLA34, YOLOv4-tiny and RetinaNet. This study can provide a reference for the research and development of mechanical harvesting equipment for Camellia oleifera and the decision-making of Camellia oleifera forest. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Harbin University of Science & Technology is the property of Journal of Harbin University of Science & Technology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Journal of South China University of Technology (Natural Science Edition) is the property of South China University of Technology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

[Objective] Vehicle detection is critical for urban intelligent transportation. Focusing on small target problems, high-density problems, and motion attribute problems, this study takes traffic surveillance images as input and aims to detect moving vehicles. [Method] Based on the anchor-free CenteNet, a detection method of multi-scale features and dual-source motion perception was proposed. Firstly, coordinate attention was introduced to the multi-scale and global context features of the network's abstraction layer, so as to supplement information in multiple stages and at multiple levels and improve the model's understanding of vehicles and scenes. Secondly, through fuzzy textures representing actual motion features of vehicles and optical flow knowledge representing general motion features of vehicles, the model's perception ability of moving vehicles was constructed. [Result] The experimental data came from the public dataset UA-DETRAC. The mean average precision (mAP) and frames per second (FPS) were used as the evaluation metrics of accuracy and speed. Experiment results show that the mAP and FPS of the proposed method are 70% and 30 frame/s respectively, which have the best balance between speed and accuracy among other compared methods. [Conclusion] It maintains that the proposed method is competent in the task of moving vehicle detection. [ABSTRACT FROM AUTHOR]

With the ever-increasing number of vehicles on the road, a faster reliable security system for university entry is needed. This paper presents an approach for Automatic Number Plate Recognition (ANPR) using deep learning and PP-OCRv3. The proposed approach utilizes a pre-trained object detection model to locate license plates, extracts a single frame of the license plate, performs license plate recognition, applies pre-processing techniques, and employs PP-OCRv3 for text extraction in real time. The system was tested with Malaysian vehicle plates, and its accuracy and speed of detection were evaluated. The results show the system's potential to be easily adapted to different camera systems, angles, and lighting conditions by retraining the deep learning model. The paper also explores various deep learning methods, such as CenterNet, EfficientDet, and Faster R-CNN, and their effectiveness in automated vehicle registration detection. The research methodology involves creating a dataset from Open Images Dataset V4, converting label text into XML files, and utilizing the TensorFlow model trained on the COCO dataset. The paper concludes with the synthetic evaluation of the trained models, comparing their performance based on precision, recall, and F1-score. Overall, the proposed approach highlights the potential of deep learning and PP-OCRv3 in achieving accurate and efficient ANPR systems.

Printed circuit board (PCB) surface defect detection is an essential part of the PCB manufacturing process. Currently, advanced CCD or CMOS sensors can capture high-resolution PCB images. However, the existing computer vision approaches for PCB surface defect detection require high computing effort, leading to insufficient efficiency. To this end, this article proposes a local and global context-enhanced lightweight CenterNet (LGCL-CenterNet) to detect PCB surface defects in real time. Specifically, we propose a two-branch lightweight vision transformer module with local and global attention, named LGT, as a complement to extract high-dimension features and leverage context-aware local enhancement after the backbone network. In the local branch, we utilize coordinate attention to aggregate more powerful features of PCB defects with different shapes. In the global branch, Bi-Level Routing Attention with pooling is used to capture long-distance pixel interactions with limited computational cost. Furthermore, a Path Aggregation Network (PANet) feature fusion structure is incorporated to mitigate the loss of shallow features caused by the increase in model depth. Then, we design a lightweight prediction head by using depthwise separable convolutions, which further compresses the computational complexity and parameters while maintaining the detection capability of the model. In the experiment, the LGCL-CenterNet increased the mAP@0.5 by 2% and 1.4%, respectively, in comparison to CenterNet-ResNet18 and YOLOv8s. Meanwhile, our approach requires fewer model parameters (0.542M) than existing techniques. The results show that the proposed method improves both detection accuracy and inference speed and indicate that the LGCL-CenterNet has better real-time performance and robustness. [ABSTRACT FROM AUTHOR]

In recent years, with the development of deep learning, SAR image-based ship target-detection technology has become a current research hotspot. SAR images are characterized by complex backgrounds, significant speckle noise interference, and poor interpretability. To address this challenge, this study proposes a new Gaussian circle radius determination strategy and a multi-level feature fusion codec-based ship detection network (MFFC-SDN) for SAR images. Differing from CenterNet's corner-oriented Gaussian circle strategy, the Gaussian circle strategy proposed in this paper takes the centre point coordinates of the real target as the origin to define a circular region. This approach allows for more precise supervision of the training model. Meanwhile, we propose MFFC-SDN to solve the problem of the poor performance of CenterNet in SAR ship detection. MFFC-SDN is an anchor-free, single-stage target-detection method with a low computation cost. In MFFC-SDN, we introduce a Multilevel Feature Recapture Codec Module (MFR-CM) to enhance the feature extraction capability of the network by using the features of the encoding stage twice because of the easy loss of image features in the lengthy codec process of the CenterNet network backbone. A Residual Attention Pyramid Feature Fusion Module (RA-PFFM) is designed to enhance the feature map and obtain multi-scale features. Experimental results on the SSDD and SAR-ship-dataset show that MFFC-SDN significantly enhances SAR image target-detection performance, effectively addressing detection challenges related to large target size variations and complex environmental conditions. Compared with existing algorithms, MFFC-SDN achieves the highest $A{P^{50}}$ A P 50 . [ABSTRACT FROM AUTHOR]

Copyright of Computer Measurement & Control is the property of Magazine Agency of Computer Measurement & Control and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The rapid development of artificial intelligence technology is gradually penetrating into multiple fields such as interior design and spatial planning. The aim of this study is to integrate artificial intelligence with interior design, enhance design artistry and user experience, and address the interactive needs of interior space design choices. A set of indoor space design recognition system has been designed by introducing artificial intelligence networks and attention mechanisms. This study first optimizes the CenterNet algorithm based on attention mechanism and feature fusion to improve its accuracy in identifying complex components. Afterwards, the long short-term memory network and convolutional neural network are trained to complete the task of spatial layout feature recognition and design. The performance test results showed that after testing 100 images, the software could recognize indoor design space images and create corresponding vector format space maps in about 5 minutes, providing them to the 3D modeling interface to generate 3D scenes. Compared to the approximately 25 minutes required by manual methods, the design efficiency has been significantly improved. The research and design method has a fast convergence speed and low loss during the retraining process. In simulation testing, its mAP value reached 91.0%, higher than similar models. It performs better in detecting walls, doors and windows, bay windows, double doors, and two-way doors. Moreover, it has outstanding ability when facing structures such as short walls and door corners, and can recognize and create vector format spatial maps within 5 minutes, which is accurate and efficient. The system designed in this project has optimized the interaction between designers and clients in interior design, accurately capturing user intentions and assisting designers in improving work efficiency. [ABSTRACT FROM AUTHOR]

The anchor-free object detection CenterNet has the problems that the utilization rate of detected object features is low, which is difficult to detect morphological changes and blurred edge objects, susceptible to interference from irrelevant information in complex backgrounds. To solve the problems above, we propose a novel anchor-free method called SDA-DLA34 in this paper. First, to solve the problem that morphological changes and blurred edge objects are difficult to detect, it is proposed that to introduce a series of deformable convolution to replace the ordinary convolution in DLA34, which effectively improve the network perception ability of morphological changes and blurred edge objects. Second, to solve the problem of low utilization of object features, it is proposed that to introduce the soft pooling layers to replace max pooling layers in the down-sampling process of DLA34, which could reduce the loss of object feature information, especially small objects. Finally, in order to pay more attention to the key information, reducing the influence of background and other irrelevant information, it is proposed to introduce attention mechanism in DLA34 to enhance the ability of the network to extract key features of the object. Experiments on MS COCO and Pascal VOC datasets have been conducted, the results show that the SDA-DLA34 is superior to to the current mainstream methods. Compared with the DLA34, the mAP, AP0.5 and AP0.75 of SDA-DLA34 increase by 8.1%, 8.0% and 6.7% respectively. [ABSTRACT FROM AUTHOR]

With the ever-increasing number of vehicles on the road, a faster reliable security system for university entry is needed. This paper presents an approach for Automatic Number Plate Recognition (ANPR) using deep learning and PP-OCRv3. The proposed approach utilizes a pre-trained object detection model to locate license plates, extracts a single frame of the license plate, performs license plate recognition, applies pre-processing techniques, and employs PP-OCRv3 for text extraction in real time. The system was tested with Malaysian vehicle plates, and its accuracy and speed of detection were evaluated. The results show the system's potential to be easily adapted to different camera systems, angles, and lighting conditions by retraining the deep learning model. The paper also explores various deep learning methods, such as CenterNet, EfficientDet, and Faster R-CNN, and their effectiveness in automated vehicle registration detection. The research methodology involves creating a dataset from Open Images Dataset V4, converting label text into XML files, and utilizing the TensorFlow model trained on the COCO dataset. The paper concludes with the synthetic evaluation of the trained models, comparing their performance based on precision, recall, and F1-score. Overall, the proposed approach highlights the potential of deep learning and PP-OCRv3 in achieving accurate and efficient ANPR systems. [ABSTRACT FROM AUTHOR]

In order to realize the intelligent recognition and statistics of hydrogenation catalyst image information, this paper presents a new method to judge the active phase by image recognition, which is different from traditional methods. Firstly, considering that hydrogenation catalyst image targets are small and easy to stack, the feature extraction network in the CenterNet model is optimized by adding the multi-feature fusion module to improve the accuracy of the network in edge positioning. Secondly, according to the linear shape of the hydrogenation catalyst, the mask branch is added to the CenterNet model to train the hydrogenation catalyst stripes with unclear target to reduce the leakage rate of the hydrogenation catalyst. The experimental results show that the detection accuracy of the improved CenterNet network is 91 % , 7 % higher than that of the original one, with a decline in detection rate by 12 % . The method proposed in this paper can accurately identify and segment the hydrogenation catalyst in the electron microscope image, which can provide technical support for the statistics and analysis of the hydrogenation catalyst image. [ABSTRACT FROM AUTHOR]

Software-Defined Networking (SDN) is a contemporary network strategy utilized instead of a traditional network structure. It provides significantly more administrative efficiency and ease than traditional networks. However, the centralized control used in SDN entails an elevated risk of single-point failure that is more susceptible to different kinds of network assaults like Distributed Denial of Service (DDoS), DoS, spoofing, and API exploitation which are very complex to identify and mitigate. Thus, a powerful intrusion detection system (IDS) based on deep learning is created in this study for the detection and mitigation of network intrusions. This system contains several stages and begins with the data augmentation method named Deep Convolutional Generative Adversarial Networks (DCGAN) to over the data imbalance problem. Then, the features are extracted from the input data using a CenterNet-based approach. After extracting effective characteristics, ResNet152V2 with Slime Mold Algorithm (SMA) based deep learning is implemented to categorize the assaults in InSDN and Edge IIoT datasets. Once the network intrusion is detected, the proposed defense module is activated to restore regular network connectivity quickly. Finally, several experiments are carried out to validate the algorithm's robustness, and the outcomes reveal that the proposed system can successfully detect and mitigate network intrusions. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Harbin University of Science & Technology is the property of Journal of Harbin University of Science & Technology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Deep learning-based visual object detection is a fundamental aspect of computer vision. These models not only locate and classify multiple objects within an image, but they also identify bounding boxes. The focus of this paper's research work is to classify fruits as ripe or overripe using digital images. Our proposed model extracts visual features from fruit images and analyzes fruit peel characteristics to predict the fruit's class. We utilize our own datasets to train two "anchor-free" models: YOLOv8 and CenterNet, aiming to produce accurate predictions. The CenterNet network primarily incorporates ResNet-50 and employs the deconvolution module DeConv for feature map upsampling. The final three branches of convolutional neural networks are applied to predict the heatmap. The YOLOv8 model leverages CSP and C2f modules for lightweight processing. After analyzing and comparing the two models, we found that the C2f module of the YOLOv8 model significantly enhances classification results, achieving an impressive accuracy rate of 99.5%. [ABSTRACT FROM AUTHOR]

Cracks on concrete surfaces are vital factors affecting construction safety. Accurate and efficient crack detection can prevent safety-related accidents. Using drones to photograph cracks on a concrete surface and detect them through computer vision technology has the advantages of accurate target recognition, simple practical operation, and low cost. To solve this problem, an improved CenterNet concrete crack-detection model is proposed. Firstly, a channel-space attention mechanism is added to the original model to enhance the ability of the convolution neural network to pay attention to the image. Secondly, a feature selection module is introduced to scale the feature map in the downsampling stage to a uniform size and combine it in the channel dimension. In the upsampling stage, the feature selection module adaptively selects the combined features and fuses them with the output features of the upsampling. Finally, the target size loss is optimized from a Smooth L1 Loss to IoU Loss to lessen its inability to adapt to targets of different sizes. The experimental results show that the improved CenterNet model reduces the FPS by 123.7 Hz, increases the GPU memory by 62 MB, increases the FLOPs by 3.81 times per second, and increases the AP by 15.4% compared with the original model. The GPU memory occupancy remained stable during the training process and exhibited good real-time performance and robustness. [ABSTRACT FROM AUTHOR]

Utilizing massive spatio-temporal sequence data and real-time synthetic aperture radar (SAR) ship target monitoring technology, it is possible to effectively predict the future trajectories and intents of ships. While real-time monitoring technology validates and adjusts spatio-temporal sequence prediction models, it still faces challenges, such as manual anchor box sizing and slow inference speeds due to large computational parameters. To address this challenge, a SAR ship target real-time detection method based on CenterNet is introduced in this article. The proposed method comprises the following steps. First, to improve the feature extraction capability of the original CenterNet network, we introduce a feature pyramid fusion structure and replace upsampled deconvolution with Deformable Convolution Networks (DCNets), which enable richer feature map outputs. Then, to identify nearshore and small target ships better, BiFormer attention mechanism and spatial pyramid pooling module are incorporated to enlarge the receptive field of network. Finally, to improve accuracy and convergence speed, we optimize the Focal loss of the heatmap and utilize Smooth L1 loss for width, height, and center point offsets, which enhance detection accuracy and generalization. Performance evaluations on two SAR image ship datasets, HRSID and SSDD, validate the method's effectiveness, achieving Average Precision (AP) values of 82.87% and 94.25%, representing improvements of 5.26% and 4.04% in AP compared to the original models, with detection speeds of 49 FPS on both datasets. These results underscore the superiority of the improved CenterNet method over other representative methods for SAR ship detection in overall performance.

The CenterNet recognition algorithm exhibits impressive performance in deep learning-based object recognition methods. Utilizing the DLA-34 as the backbone for object detection, it achieves higher average precision than one-stage algorithms, albeit at a slower detection speed. We propose an enhanced recognition algorithm named FPN-CenterNet to address this problem, which leverages the Feature Pyramid Networks (FPN) to refine the backbone. The goal is to maintain accuracy while improving detection speed, making it more suitable for real-time 3D reconstruction tasks. Specifically, this approach adopts a feature pyramid structure to extract features at multiple scales, promptly utilizing these features across different levels for reinforcement. The original DLA-34 network, which relies on more modules and parameters for inter-scale feature interaction, reduces recognition efficiency. The proposed algorithm first employs the improved object recognition network to detect objects within an image. Subsequently, the recognition network identifies the objects and performs convolutional operations to extract their depth features. Finally, these extracted depth features are used to render the pre-modeled models of the objects, producing the desired output. 3D reconstruction is an application of recognition results in our experiment. Experimental evaluations on the Dunhuang Buddhist Art Image dataset involve comparing the proposed method and eight existing image recognition algorithms. The assessment encompasses both recognition speed and accuracy. The results demonstrate the outstanding performance of the proposed recognition method, achieving both swift recognition and facilitating 3D reconstruction of objects.

Malaria is a blood disease caused by the Plasmodium parasite that is transmitted through the bite of female Anopheles mosquitoes. These mosquitoes can cross borders without passports or visas, making malaria a global health concern. To effectively treat malaria, infectious disease specialists must monitor the efficacy of the treatment by counting the number of parasites in a patient’s blood at various time intervals. However, this task is challenging because it involves examining thin or thick blood smear samples under a microscope, which can be tiring to the human eye, particularly when there are many infected patients or when there is a shortage of clinical experts. In such cases, rapid diagnosis is crucial. One approach is to capture microscopic images of blood smear samples using a camera and then employ deep learning-based object detection models to detect and count the infected red blood cells. In this study, state-of-the-art object detection models, including CenterNet, EfficientDet, Faster R-CNN, RetinaNet, and YOLOv8, were explored. The dataset was generated using thin blood smear images in the laboratory. The results revealed that YOLOv8s outperformed the other models, achieving an score of 0.9031 and an mAP@[0.50:0.05:0.95] score of 0.5957. This study also found that various model combinations and ensemble strategies could improve the detection of malaria parasites. Specifically, the weighted boxes fusion ensembling approach achieved an score of 0.9186 and an mAP@[0.50:0.05:0.95] score of 0.6196. In contrast, the non-maximum weighted method achieved an score of 0.9324 and an mAP@[0.50:0.05:0.95] score of 0.6214.

Synthetic aperture radar (SAR) target detection plays a crucial role in both military and civilian fields, attracting significant attention from researchers globally. CenterNet, a single-stage target detection method, is known for its high detection speed and accuracy by eliminating anchor-related calculations and nonmaximum suppression. However, directly applying CenterNet to SAR ship detection poses challenges due to the distinctive characteristics of SAR images, including lower resolution, lower signal-to-noise ratio, and larger ship aspect ratios. To address these challenges, we propose MPDNet. which introduces a multilevel pyramid feature extraction module (MP-FEM) to replace the encoding–decoding structure in CenterNet. MP-FEM employs multilevel pyramid and channel compression to fuse multiscale SAR image features and acquire deep features quickly. Second, we propose the convolution channel attention module, which improves the multilayer perceptron in the common pooling attention mechanism into a multistage and 1-D convolution. Therefore, the feature extraction capability of MP-FEM is further refined. Furthermore, we propose the detection task decoupling module (DTDM), which considers the characteristics of SAR ships and effectively detects smaller targets of different sizes, distinguishing the centers and sizes of densely arranged ships. DTDM extracts task-related features from the original feature map before inputting it into the three detection headers, thereby addressing the problem of task coupling in CenterNet's detection header module for SAR ship detection. Finally, the experimental results on SSDD dataset and SAR-ship-dataset show that the proposed network can significantly improve the SAR target detection accuracy.

This paper proposes a novel object detection method to address the challenges posed by small objects and occlusion in object detection. This work is performed within the CenterNet framework, leveraging the MobileNetV3 backbone to model the input image’s abstract representation in a lightweight manner. A sparse convolutional skip connection is introduced in the bottleneck of MobileNetV3, specifically designed to adaptively suppress redundant and interfering information, thus enhancing feature extraction capabilities. A Dual-Path Bidirectional Feature Pyramid Network (DBi-FPN) is incorporated, allowing for high-level feature fusion through bidirectional flow and significantly improving the detection capabilities for small objects and occlusions. Task heads are applied within the feature space of multi-scale information merged by DBi-FPN, facilitating comprehensive consideration of multi-level representations. A bounding box-area loss function is also introduced, aimed at enhancing the model’s adaptability to object morphologies and geometric distortions. Extensive experiments on the PASCAL VOC 2007 and MS COCO 2017 datasets validate the competitiveness of our proposed method, particularly in real-time applications on resource-constrained devices. Our contributions offer promising avenues for enhancing the accuracy and robustness of object detection systems in complex scenarios.

This paper proposes a real-time detection method to improve the Infrared small target detection CenterNet (ISTDCenterNet) network for detecting small infrared targets in complex environments. Themethod eliminates the need for an anchor frame, addressing the issues of low accuracy and slow speed. HRNet is used as the framework for feature extraction, and an ECBAM attention module is added to each stage branch for intelligent identification of the positions of small targets and significant objects. A scale enhancement module is also added to obtain a high-level semantic representation and fine-resolution prediction map for the entire infrared image. Besides, an improved sensory field enhancementmodule is designed to leverage semantic information in low-resolution feature maps, and a convolutional attention mechanism module is used to increase network stability and convergence speed. Comparison experiments conducted on the infrared small target data set ESIRST. The experiments show that compared to the benchmark network CenterNet-HRNet, the proposed ISTD-CenterNet improves the recall by 22.85% and the detection accuracy by 13.36%. Compared to the state-of-the-art YOLOv5small, the ISTDCenterNet recall is improved by 5.88%, the detection precision is improved by 2.33%, and the detection frame rate is 48.94 frames/sec, which realizes the accurate real-time detection of small infrared targets. [ABSTRACT FROM AUTHOR]

Data Matrix (DM) codes have been widely used in industrial production. The reading of DM code usually includes positioning and decoding. Accurate positioning is a prerequisite for successful decoding. Traditional image processing methods have poor adaptability to pollution and complex backgrounds. Although deep learning-based methods can automatically extract features, the bounding boxes cannot entirely fit the contour of the code. Further image processing methods are required for precise positioning, which will reduce efficiency. Because of the above problems, a CenterNet-based DM code key point detection network is proposed, which can directly obtain the four key points of the DM code. Compared with the existing methods, the degree of fitness is higher, which is conducive to direct decoding. To further improve the positioning accuracy, an enhanced loss function is designed, including DM code key point heatmap loss, standard DM code projection loss, and polygon Intersection-over-Union (IoU) loss, which is beneficial for the network to learn the spatial geometric characteristics of DM code. The experiment is carried out on the self-made DM code key point detection dataset, including pollution, complex background, small objects, etc., which uses the Average Precision (AP) of the common object detection metric as the evaluation metric. AP reaches 95.80%, and Frames Per Second (FPS) gets 88.12 on the test set of the proposed dataset, which can achieve real-time performance in practical applications. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Remote Sensing is the property of Editorial Office of Journal of Remote Sensing & Science Publishing Co. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The overgrowth of weeds growing along with the primary crop in the fields reduces crop production. Conventional solutions like hand weeding are labor-intensive, costly, and time-consuming; farmers have used herbicides. The application of herbicide is effective but causes environmental and health concerns. Hence, Precision Agriculture (PA) suggests the variable spraying of herbicides so that herbicide chemicals do not affect the primary plants. Motivated by the gap above, we proposed a Deep Learning (DL) based model for detecting Eggplant (Brinjal) weed in this paper. The key objective of this study is to detect plant and non-plant (weed) parts from crop images. With the help of object detection, the precise location of weeds from images can be achieved. The dataset is collected manually from a private farm in Gandhinagar, Gujarat, India. The combined approach of classification and object detection is applied in the proposed model. The Convolutional Neural Network (CNN) model is used to classify weed and non-weed images; further DL models are applied for object detection. We have compared DL models based on accuracy, memory usage, and Intersection over Union (IoU). ResNet-18, YOLOv3, CenterNet, and Faster RCNN are used in the proposed work. CenterNet outperforms all other models in terms of accuracy, i.e., 88%. Compared to other models, YOLOv3 is the least memory-intensive, utilizing 4.78 GB to evaluate the data. [ABSTRACT FROM AUTHOR]

Abstract In this paper, the authors apply the feature pyramid network (FPN) to the single‐stage anchor‐free object detection algorithm CenterNet, and the effectiveness of the multi‐level feature fusion of FPN for the object detection algorithm is proved by experiments. However, multi‐level feature fusion leads to an increase in computational cost. In this regard, this paper proposes an object detection algorithm, called DDE‐Net, that does not use multi‐level feature fusion and only uses single‐level feature for optimization. The key component in it: the dense dilated encoder, which encourages dense information exchange of features between different spatial scales. This paper presents extensive experiments, and DDE‐Net shows strong performance compared to that of other popular models on the PASCAL VOC and on the COCO2017 dataset. On the COCO2017 dataset, the authors’ DDE‐Net achieves comparable results with its feature pyramids counterpart RetinaNet, while applying the same backbone with smaller params and GFLOPs than RetinaNet. With an image size of 512 × 512, DDE‐Net achieves 37.3 AP running at 81 fps on 2080 Ti.

ObjectivePineapple is a common tropical fruit, and its ripeness has an important impact on the storage and marketing. It is particularly important to analyze the maturity of pineapple fruit before picking. Deep learning technology can be an effective method to achieve automatic recognition of pineapple maturity. To improve the accuracy and rate of automatic recognition of pineapple maturity, a new network model named MobileNet V3-YOLOv4 was proposed in this study.MethodsFirstly, pineapple maturity analysis data set was constructed. A total of 1580 images were obtained, with 1264 images selected as the training set, 158 images as the validation set, and 158 images as the test set. Pineapple photos were taken in natural environment. In order to ensure the diversity of the data set and improve the robustness and generalization of the network, pineapple photos were taken under the influence of different factors such as branches and leaves occlusion, uneven lighting, overlapping shadows, etc. and the location, weather and growing environment of the collection were different. Then, according to the maturity index of pineapple, the photos of pineapple with different maturity were marked, and the labels were divided into yellow ripeness and green ripeness. The annotated images were taken as data sets and input into the network for training. Aiming at the problems of the traditional YOLOv4 network, such as large number of parameters, complex network structure and slow reasoning speed, a more optimized lightweight MobileNet V3-YOLOv4 network model was proposed. The model utilizes the benck structure to replace the Resblock in the CSPDarknet backbone network of YOLOv4. Meanwhile, in order to verify the effectiveness of the MobileNet V3-YOLOv4 network, MobileNet V1-YOLOv4 model and MobileNet V2-YOLOv4 model were also trained. Five different single-stage and two-stage network models, including R-CNN, YOLOv3, SSD300, Retinanet and Centernet were compared with each evaluation index to analyze the performance superiority of MobileNet V3-YOLOv4 model.Results and Discussions]MobileNet V3-YOLOv4 was validated for its effectiveness in pineapple maturity detection through experiments comparing model performance, model classification prediction, and accuracy tests in complex pineapple detection environments.The experimental results show that, in terms of model performance comparison, the training time of MobileNet V3-YOLOv4 was 11,924 s, with an average training time of 39.75 s per round, the number of parameters was 53.7 MB, resulting in a 25.59% reduction in the saturation time compared to YOLOv4, and the parameter count accounted for only 22%. The mean average precision (mAP) of the trained MobileNet V3-YOLOv4 in the verification set was 53.7 MB. In order to validate the classification prediction performance of the MobileNet V3-YOLOv4 model, four metrics, including Recall score, F1 Score, Precision, and average precision (AP), were utilized to classify and recognize pineapples of different maturities. The experimental results demonstrate that MobileNet V3-YOLOv4 exhibited significantly higher Precision, AP, and F1 Score the other. For the semi-ripe stage, there was a 4.49% increase in AP, 0.07 improvement in F1 Score, 1% increase in Recall, and 3.34% increase in Precision than YOLOv4. As for the ripe stage, there was a 6.06% increase in AP, 0.13 improvement in F1 Score, 16.55% increase in Recall, and 6.25% increase in Precision. Due to the distinct color features of ripe pineapples and their easy differentiation from the background, the improved network achieved a precision rate of 100.00%. Additionally, the mAP and reasoning speed (Frames Per Second, FPS) of nine algorithms were examined. The results showed that MobileNet V3-YOLOv4 achieved an mAP of 90.92%, which was 5.28% higher than YOLOv4 and 3.67% higher than YOLOv3. The FPS was measured at 80.85 img/s, which was 40.28 img/s higher than YOLOv4 and 8.91 img/s higher than SSD300. The detection results of MobileNet V3-YOLOv4 for pineapples of different maturities in complex environments indicated a 100% success rate for both the semi-ripe and ripe stages, while YOLOv4, MobileNet V1-YOLOv4, and MobileNet V2-YOLOv4 exhibited varying degrees of missed detections.ConclusionsBased on the above experimental results, it can be concluded that MobileNet V3-YOLOv4 proposed in this study could not only reduce the training speed and parameter number number, but also improve the accuracy and reasoning speed of pineapple maturity recognition, so it has important application prospects in the field of smart orchard. At the same time, the pineapple photo data set collected in this research can also provide valuable data resources for the research and application of related fields.

A belt conveyor system is one of the essential equipment in coal mining. The damages to conveyor belts are hazardous because they would affect the stable operation of a belt conveyor system whilst impairing the coal mining efficiency. To address these problems, a novel conveyor belt damage detection method based on CenterNet is proposed in this paper. The fusion of feature-wise and response-wise knowledge distillation is proposed, which balances the performance and size of the proposed deep neural network. The Fused Channel-Spatial Attention is proposed to compress the latent feature maps efficiently, and the Kullback-Leibler divergence is introduced to minimize the distribution distance between student and teacher networks. Experimental results show that the proposed lightweight object detection model reaches 92.53% mAP and 65.8 FPS. The proposed belt damage detection system can detect conveyor belt damages efficiently and accurately, which indicates its high potential to deploy on end devices.

Printed circuit board (PCB) surface defect detection is an essential part of the PCB manufacturing process. Currently, advanced CCD or CMOS sensors can capture high-resolution PCB images. However, the existing computer vision approaches for PCB surface defect detection require high computing effort, leading to insufficient efficiency. To this end, this article proposes a local and global context-enhanced lightweight CenterNet (LGCL-CenterNet) to detect PCB surface defects in real time. Specifically, we propose a two-branch lightweight vision transformer module with local and global attention, named LGT, as a complement to extract high-dimension features and leverage context-aware local enhancement after the backbone network. In the local branch, we utilize coordinate attention to aggregate more powerful features of PCB defects with different shapes. In the global branch, Bi-Level Routing Attention with pooling is used to capture long-distance pixel interactions with limited computational cost. Furthermore, a Path Aggregation Network (PANet) feature fusion structure is incorporated to mitigate the loss of shallow features caused by the increase in model depth. Then, we design a lightweight prediction head by using depthwise separable convolutions, which further compresses the computational complexity and parameters while maintaining the detection capability of the model. In the experiment, the LGCL-CenterNet increased the mAP@0.5 by 2% and 1.4%, respectively, in comparison to CenterNet-ResNet18 and YOLOv8s. Meanwhile, our approach requires fewer model parameters (0.542M) than existing techniques. The results show that the proposed method improves both detection accuracy and inference speed and indicate that the LGCL-CenterNet has better real-time performance and robustness.

Sooty mold is a common disease found in citrus plants and is characterized by black fungi growth on fruits, leaves, and branches. This mold reduces the plant's ability to carry out photosynthesis. In small leaves, it is very difficult to detect sooty mold at the early stages. Deep learning-based image recognition techniques have the potential to identify and diagnose pest damage and diseases such as sooty mold. Recent studies used advanced and expensive hyperspectral or multispectral cameras attached to UAVs to examine the canopy of the plants and mid-range cameras to capture close-up infected leaf images. To bridge the gap on capturing canopy level images using affordable camera sensors, this study used a low-cost home surveillance camera to monitor and detect sooty mold infection on citrus canopy combined with deep learning algorithms. To overcome the challenges posed by varying light conditions, the main reason for using specialized cameras, images were collected at night, utilizing the camera's built-in night vision feature. A total of 4200 sliced night-captured images were used for training, 200 for validation, and 100 for testing, employed on the YOLOv5m, YOLOv7, and CenterNet models for comparison. The results showed that YOLOv7 was the most accurate in detecting sooty molds at night, with 74.4% mAP compared to YOLOv5m (72%) and CenterNet (70.3%). The models were also tested using preprocessed (unsliced) night images and day-captured sliced and unsliced images. The testing on preprocessed (unsliced) night images demonstrated the same trend as the training results, with YOLOv7 performing best compared to YOLOv5m and CenterNet. In contrast, testing on the day-captured images had underwhelming outcomes for both sliced and unsliced images. In general, YOLOv7 performed best in detecting sooty mold infections at night on citrus canopy and showed promising potential in real-time orchard disease monitoring and detection. Moreover, this study demonstrated that utilizing a cost-effective surveillance camera and deep learning algorithms can accurately detect sooty molds at night, enabling growers to effectively monitor and identify occurrences of the disease at the canopy level. [ABSTRACT FROM AUTHOR]

Safety helmet detection is a hot topic of research in the field of industrial safety for object detection technology. Existing object detection methods still face great challenges for the detection of small-scale safety helmet object. In this paper, we propose a safety helmet detection method based on the fusion of semantic guidance and feature selection. The method is able to consider the balance between detection performance and efficiency. First, a multi-scale non-local module is proposed to establish internal correlations between different scales of deep image features as well as to aggregate semantic context information to guide the information recovery of decoder network features. Then the feature selection fusion structure is proposed to adaptively select deep features and underlying key features for fusion to make up for the missing semantic and spatial detail information of the decoding network and improve the spatial location expression capability of the decoding network. Experimental analysis shows that the method in this paper has good detection performance on the expanded safety helmet wearing dataset with 5.12% improvement in mAP compared to the baseline method CenterNet, and 6.11% improvement in AP for the safety helmet object. [ABSTRACT FROM AUTHOR]

The detection and accurate positioning of agricultural pests and diseases can significantly improve the effectiveness of disease and pest control and reduce the cost of prevention and control, which has become an urgent need for crop production. Aiming at the low precision of maize leaf pest and disease detection, a new model of maize leaf pest and disease detection using attention mechanism and multi-scale features was proposed. Our model combines a convolutional block attention module (CBAM) with the ResNet50 backbone network to suppress complex background interference and enhance feature expression in specific regions of the maize leaf images. We also design a multi-scale feature fusion module that aggregates local and global information at different scales, improving the detection performance for objects of varying sizes. This module reduces the number of parameters and enhances efficiency by using a lightweight module and replacing the deconvolutional layer. Experimental results on a natural environment dataset demonstrate that our proposed model achieves an average detection accuracy of 85.13%, which is 9.59% higher than the original CenterNet model. The model has 24.296 M parameters and a detection speed of 23.69 f/s. Compared with other popular models such as SSD-VGG, YOLOv5, Faster-RCNN, and Efficientdet-D0, our proposed model demonstrates superior performance in the fast and accurate detection of maize leaf pests and diseases. This model has practical applications in the identification and treatment of maize pests and diseases in the field, and it can provide technical support for precision pesticide application. The trained model can be deployed to a web client for user convenience. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)