Detection Speed Research Articles

DFFNet: a lightweight approach for efficient feature-optimized fusion in steel strip surface defect detection

Steel surface defect detection is crucial in manufacturing, but achieving high accuracy and real-time performance with limited computing resources is challenging. To address this issue, this paper proposes DFFNet, a lightweight fusion network, for fast and accurate steel surface defect detection. Firstly, a lightweight backbone network called LDD is introduced, utilizing partial convolution to reduce computational complexity and extract spatial features efficiently. Then, PANet is enhanced using the Efficient Feature-Optimized Converged Network and a Feature Enhancement Aggregation Module (FEAM) to improve feature fusion. FEAM combines the Efficient Layer Aggregation Network and reparameterization techniques to extend the receptive field for defect perception, and reduce information loss for small defects. Finally, a WIOU loss function with a dynamic non-monotonic mechanism is designed to improve defect localization in complex scenes. Evaluation results on the NEU-DET dataset demonstrate that the proposed DFFNet achieves competitive accuracy with lower computational complexity, with a detection speed of 101 FPS, meeting real-time performance requirements in industrial settings. Furthermore, experimental results on the PASCAL VOC and MS COCO datasets demonstrate the strong generalization capability of DFFNet for object detection in diverse scenarios.

Steel Surface Defect Detection Algorithm Based on Improved YOLOv8n

The traditional detection methods of steel surface defects have some problems, such as a lack of feature extraction ability, sluggish detection speed, and subpar detection performance. In this paper, a YOLOv8-based DDI-YOLO model is suggested for effective steel surface defect detection. First, on the Backbone network, the extended residual module (DWR) is fused with the C2f module to obtain C2f_DWR, and the two-step approach is used to carry out the effective extraction of multiscale contextual information, and then fusing feature maps formed from the multiscale receptive fields to enhance the capacity for feature extraction. Also based on the above, an extended heavy parameter module (DRB) is added to the structure of C2f_DWR to make up for the lack of C2f’s ability to capture small-scale pattern defects between training to enhance the training fluency of the model. Finally, the Inner-IoU loss function is employed to enhance the regression accuracy and training speed of the model. The experimental results show that the detection of DDI-YOLO on the NEU-DET dataset improves the mAP by 2.4%, the accuracy by 3.3%, and the FPS by 59 frames/s compared with the original YOLOv8n.Therefore, this paper’s proposed model has a superior mAP, accuracy, and FPS in identifying surface defects in steel.

A lightweight fire detection algorithm for small targets based on YOLOv5s

In response to the current challenges fire detection algorithms encounter, including low detection accuracy and limited recognition rates for small fire targets in complex environments, we present a lightweight fire detection algorithm based on an improved YOLOv5s. The introduction of the CoT (Contextual Transformer) structure into the backbone neural network, along with the creation of the novel CSP1_CoT (Cross stage partial 1_contextual transformer) module, has effectively reduced the model’s parameter count while simultaneously enhancing the feature extraction and fusion capabilities of the backbone network; The network’s Neck architecture has been extended by introducing a dedicated detection layer tailored for small targets and incorporating the SE (Squeeze-and-Excitation) attention mechanism. This augmentation, while minimizing parameter proliferation, has significantly bolstered the interaction of multi-feature information, resulting in an enhanced small target detection capability; The substitution of the original loss function with the Focal-EIoU (Focal-Efficient IoU) loss function has yielded a further improvement in the model’s convergence speed and precision; The experimental results indicate that the modified model achieves an mAP@.5 of 96% and an accuracy of 94.8%, marking improvements of 8.8% and 8.9%, respectively, over the original model. Furthermore, the model’s parameter count has been reduced by 1.1%, resulting in a compact model size of only 14.6MB. Additionally, the detection speed has reached 85 FPS (Frames Per Second), thus satisfying real-time detection requirements. This enhancement in precision and accuracy, while simultaneously meeting real-time and lightweight constraints, effectively caters to the demands of fire detection.

Multiple object characterization of recyclable domestic waste using binocular stereo vision

Recycling and recovery of domestic waste pose significant challenges in terms of accurate object positioning and ranging, areas where conventional methods, heavily reliant on monocular image processing, often fail due to limited distance data and reduced precision. This study introduces an innovative strategy for the localization and ranging of recyclable waste objects, leveraging the NanoDet-Plus model within a binocular stereo vision context. A dataset, the Binocular Recyclable Domestic Waste Dataset (BRDWD), was constructed, with transfer learning executed via the pretrained MULTI-TRASH monocular dataset. Procedures encompassed stereo camera calibration and rectification to address geometric distortions, while the integration of the Semi-Global Block Matching (SGBM) algorithm supplied disparity details for triangulation-based distance computations. Enhancements to the NanoDet-Plus model led to improved target accuracy, evidenced by a mean average precision at a threshold of 0.5 (mAP@0.5) reaching 91.88% and a detection speed of 15 frames per second (FPS). Assessments verified a ranging error below 5% within the 0.5 to 1.2-meter span, aligning well with the requirements of typical recyclable waste sorting and recycling situations. Additionally, a human-machine interface was developed utilizing Pyside6, enabling image uploads, real-time result visualization, and interactive process control, thereby furnishing pivotal advancements for the automation and intellectualization of waste categorization and recycling.

Harmonic Detection Methods and Optimization Strategies for Charger in Power Grid

The charger’s operation in the power grid will cause harmonic pollution, affecting the service life of the equipment and the quality of the power grid. The harmonic characteristics and detection during the charging process are studied. The simulation model of charger is established first using MATLAB, the results of the experiment show that odd harmonics are mainly generated during charging, and the distortion rate of the current harmonics remains between 0.231 and 0.443 during the charging process. A new wavelet neural network is proposed to detect the harmonic method. Experimental results indicate that, compared with previous detection methods, this network has better detection efficiency and accuracy for harmonics. But detection result of this network is unstable and does not even converge. A new adaptive bat algorithm with improved focusing distance is presented here to optimize the wavelet neural network, which effectively solves the defect of being sensitive to the node number of hidden layer, resulting in higher detection accuracy, faster detection speed, and stable convergence. The detection error range is reduced from −0.03 to 0.02, and the number of iterations is reduced to around 1000.

WH-DETR: An Efficient Network Architecture for Wheat Spike Detection in Complex Backgrounds

Wheat spike detection is crucial for estimating wheat yields and has a significant impact on the modernization of wheat cultivation and the advancement of precision agriculture. This study explores the application of the DETR (Detection Transformer) architecture in wheat spike detection, introducing a new perspective to this task. We propose a high-precision end-to-end network named WH-DETR, which is based on an enhanced RT-DETR architecture. Initially, we employ data augmentation techniques such as image rotation, scaling, and random occlusion on the GWHD2021 dataset to improve the model’s generalization across various scenarios. A lightweight feature pyramid, GS-BiFPN, is implemented in the network’s neck section to effectively extract the multi-scale features of wheat spikes in complex environments, such as those with occlusions, overlaps, and extreme lighting conditions. Additionally, the introduction of GSConv enhances the network precision while reducing the computational costs, thereby controlling the detection speed. Furthermore, the EIoU metric is integrated into the loss function, refined to better focus on partially occluded or overlapping spikes. The testing results on the dataset demonstrate that this method achieves an Average Precision (AP) of 95.7%, surpassing current state-of-the-art object detection methods in both precision and speed. These findings confirm that our approach more closely meets the practical requirements for wheat spike detection compared to existing methods.

A Lightweight Pine Wilt Disease Detection Method Based on Vision Transformer-Enhanced YOLO

Pine wilt disease (PWD) is a forest disease characterized by rapid spread and extremely high lethality, posing a serious threat to the ecological security of China’s forests and causing significant economic losses in forestry. Given the extensive forestry area, limited personnel for inspection and monitoring, and high costs, utilizing UAV-based remote sensing monitoring for diseased trees represents an effective approach for controlling the spread of PWD. However, due to the small target size and uneven scale of pine wilt disease, as well as the limitations of real-time detection by drones, traditional disease tree detection algorithms based on RGB remote sensing images do not achieve an optimal balance among accuracy, detection speed, and model complexity due to real-time detection limitations. Consequently, this paper proposes Light-ViTeYOLO, a lightweight pine wilt disease detection method based on Vision Transformer-enhanced YOLO (You Only Look Once). A novel lightweight multi-scale attention module is introduced to construct an EfficientViT feature extraction network for global receptive field and multi-scale learning. A novel neck network, CACSNet(Content-Aware Cross-Scale bidirectional fusion neck network), is designed to enhance the detection of diseased trees at single granularity, and the loss function is optimized to improve localization accuracy. The algorithm effectively reduces the number of parameters and giga floating-point operations per second (GFLOPs) of the detection model while enhancing overall detection performance. Experimental results demonstrate that compared with other baseline algorithms, Light-ViTeYOLO proposed in this paper has the least parameter and computational complexity among related algorithms, with 3.89 MFLOPs and 7.4 GFLOPs, respectively. The FPS rate is 57.9 (frames/s), which is better than the original YOLOv5. Meanwhile, its mAP@0.5:0.95 is the best among the baseline algorithms, and the recall and mAP@0.5 slightly decrease. Our Light-ViTeYOLO is the first lightweight method specifically designed for detecting pine wilt disease. It not only meets the requirements for real-time detection of pine wilt disease outbreaks but also provides strong technical support for automated forestry work.

Loop Closure Detection with CNN in RGB-D SLAM for Intelligent Agricultural Equipment

Loop closure detection plays an important role in the construction of reliable maps for intelligent agricultural machinery equipment. With the combination of convolutional neural networks (CNN), its accuracy and real-time performance are better than those based on traditional manual features. However, due to the use of small embedded devices in agricultural machinery and the need to handle multiple tasks simultaneously, achieving optimal response speeds becomes challenging, especially when operating on large networks. This emphasizes the need to study in depth the kind of lightweight CNN loop closure detection algorithm more suitable for intelligent agricultural machinery. This paper compares a variety of loop closure detection based on lightweight CNN features. Specifically, we prove that GhostNet with feature reuse can extract image features with both high-dimensional semantic information and low-dimensional geometric information, which can significantly improve the loop closure detection accuracy and real-time performance. To further enhance the speed of detection, we implement Multi-Probe Random Hyperplane Local Sensitive Hashing (LSH) algorithms. We evaluate our approach using both a public dataset and a proprietary greenhouse dataset, employing an incremental data processing method. The results demonstrate that GhostNet and the Linear Scanning Multi-Probe LSH algorithm synergize to meet the precision and real-time requirements of agricultural closed-loop detection.

100-kHz carbon monoxide TP-PLIF imaging using ultra-narrow-linewidth burst-mode OPO.

A 100-kHz rate two-photon planar laser-induced fluorescence (TP-PLIF) imaging of carbon monoxide (CO) is successfully demonstrated utilizing a narrow-linewidth optical parametric oscillator (OPO) generating light at ∼230.1 nm. A specially designed injection-seeded burst-mode OPO was constructed and characterized for this purpose. This OPO efficiently converts the 355-nm output of a high-energy nanosecond burst-mode laser to ∼230.1 nm following parametric splitting and mixing processes. Generation of an ultra-narrow-linewidth 230.1 nm laser pulse is crucial for effectively exciting CO via a two-photon process from the ground X1Σ+ to the B1Σ+ electronic state-enabling PLIF imaging over a large area. The experimental setup is capable of tracking high-speed flow structures of a CO/N2 mixture, showcasing detection speeds 100 times greater than those achieved with previous femtosecond laser sources. This substantial increase in repetition rate will allow time-resolved CO-TP-PLIF measurements in highly dynamic hypersonic boundary layers and detonation-driven combustion processes for revealing chemical kinetics and turbulent aerodynamics.

Detection Strategies for Volatile Fragrance Released from Agricultural Products: Progress and Prospects

AbstractThe volatile aroma released from agricultural products is closely related to the health status and quality of their growth, thus endowing the related detection with great significance. For example, the dynamic variation of the volatile chemical composition of a banana during the growth process can reflect its ripeness. Also important for quality monitoring and storage is to precisely and swiftly identify volatile compounds produced by mildew in rice and wheat. In this endeavor, the current detection technologies such as gas chromatography‐mass spectrometry method (GC‐MS) cannot meet the pressing needs of smart agriculture in terms of real‐time monitoring, cost‐effectiveness, sensitivity, and detection speed, thereby necessitating alternative strategies to simultaneously satisfy these requirements. Aiming to provide an overall development trend in this field, this paper summarizes the existing detection technologies including GC‐MS, E‐nose, and sensory analysis with their respective shortcomings and challenges, and then proposes the application prospects. This work can effectively enrich the reliable monitoring methods for volatile agricultural fragrance while promoting the long‐run development of smart agriculture.

IoT-Enabled Single-Camera Speed Sensor for Smart City Tasks

This article presents an innovative IoT-enabled single-camera speed sensor designed for smart city applications. The research encompasses the development of both hardware and software components, focusing on a computer vision and artificial neural network-based system. A novel aspect of this study is the implementation of a distance-measuring algorithm that eliminates the need for expensive LIDAR sensors traditionally used in speed cameras. Instead, the system relies on convolutional neural networks (CNN) and computer vision algorithms to estimate vehicle speed accurately. Field testing was conducted in real conditions over several months, generating sufficient data to assess the device’s ability to function under various adverse conditions. The results demonstrate the system’s capability to perform vehicle speed detection consistently across diverse conditions, showcasing its potential as a scalable solution for urban traffic management. This makes the proposed system not only more affordable but also simpler to deploy and maintain, thereby enhancing its suitability for widespread use in smart city environments.

System Development and Assessment For Road Vehicles Speed Detection Using GSM

System Development and Assessment For Road Vehicles Speed Detection Using GSM

Advances in Nanoplasmonic Biosensors: Optimizing Performance for Exosome Detection Applications.

The development of sensitive and specific exosome detection tools is essential because they are believed to provide specific information that is important for early detection, screening, diagnosis, and monitoring of cancer. Among the many detection tools, surface-plasmon resonance (SPR) biosensors are analytical devices that offer advantages in sensitivity and detection speed, thereby making the sample-analysis process faster and more accurate. In addition, the penetration depth of the SPR biosensor, which is <300 nm, is comparable to the size of the exosome, making the SPR biosensor ideal for use in exosome research. On the other hand, another type of nanoplasmonic sensor, namely a localized surface-plasmon resonance (LSPR) biosensor, has a shorter penetration depth of around 6 nm. Structural optimization through the addition of supporting layers and gap control between particles is needed to strengthen the surface-plasmon field. This paper summarizes the progress of the development of SPR and LSPR biosensors for detecting exosomes. Techniques in signal amplification from two sensors will be discussed. There are three main parts to this paper. The first two parts will focus on reviewing the working principles of each sensor and introducing several methods that can be used to isolate exosomes. This article will close by explaining the various sensor systems that have been developed and the optimizations carried out to obtain sensors with better performance. To illustrate the performance improvements in each sensor system discussed, the parameters highlighted include the detection limit, dynamic range, and sensitivity.

A novel robotic-assisted deep learning-enabled computer vision approach for nondestructive diagnosis of railway bolt faults

Abstract Railways play a vital role in the inland transportation system worldwide, and abnormal bolt components at the track joints are the main cause of train accidents. The detection and identification of faults in rail bolt components are of considerable research importance. To address this problem, numerous researchers have opted for computer vision-based methods to accomplish the detection and identification of the target, but the existing methods have poor detection performance diminished detection capabilities when the target position changes or some feature information is occluded, and the detection speed and accuracy are far from meeting the requirements of practical applications. Therefore, based on the construction of a dedicated dataset for bolt components, this paper uses the K-means dimensional clustering algorithm to re-cluster the dataset according to the target size characteristics, with the aim of reduce the bounding box regression error. At the same time, a novel loss function iteration method is proposed by incorporating an adaptive optimization algorithm, in order to improve the detection speed and ensure good convergence, and the model complexity is reduced based on deep model pruning. Finally, the optimized detection model is implemented on the robotic-assisted platform for testing, and the experimental results verify that the algorithm can quickly and accurately complete various fault diagnosis tasks of bolt components in practical applications. The main achievements of this study include the construction of a large-scale image dataset for novel rail bolt components and propelled the application of deep learning methods in vision-based rail bolt fault diagnosis.

Real-Time Object Detection from UAV Inspection Videos by Combining YOLOv5s and DeepStream.

The high-altitude real-time inspection of unmanned aerial vehicles (UAVs) has always been a very challenging task. Because high-altitude inspections are susceptible to interference from different weather conditions, interference from communication signals and a larger field of view result in a smaller object area to be identified. We adopted a method that combines a UAV system scheduling platform with artificial intelligence object detection to implement the UAV automatic inspection technology. We trained the YOLOv5s model on five different categories of vehicle data sets, in which mAP50 and mAP50-95 reached 93.2% and 71.7%, respectively. The YOLOv5s model size is only 13.76 MB, and the detection speed of a single inspection photo reaches 11.26 ms. It is a relatively lightweight model and is suitable for deployment on edge devices for real-time detection. In the original DeepStream framework, we set up the http communication protocol to start quickly to enable different users to call and use it at the same time. In addition, asynchronous sending of alarm frame interception function was added and the auxiliary services were set up to quickly resume video streaming after interruption. We deployed the trained YOLOv5s model on the improved DeepStream framework to implement automatic UAV inspection.

DiffRank: Enhancing efficiency in discontinuous frame rate analysis for urban surveillance systems

Urban public safety management relies heavily on video surveillance systems, which provide crucial visual data for resolving a wide range of incidents and controlling unlawful activities. Traditional methods for target detection predominantly employ a two-stage approach, focusing on precision in identifying objects such as pedestrians and vehicles. These objects, typically sparse in large-scale, lower-quality surveillance footage, induce considerable redundant computation during the initial processing stage. This redundancy constrains real-time detection capabilities and escalates processing costs. Furthermore, transmitting raw images and videos laden with superfluous information to centralized back-end systems significantly burdens network communications and fails to capitalize on the computational resources available at diverse surveillance nodes. This study introduces DiffRank, a novel preprocessing method for fixed-angle video imagery in urban surveillance. The method strategically generates candidate regions during preprocessing, thereby reducing redundant object detection and improving the efficiency of the detection algorithm. Drawing upon change detection principles, a background feature learning approach utilizing shallow features has been developed. This approach prioritizes learning the characteristics of fixed-area backgrounds over direct background identification. As a result, alterations in ROI are efficiently discerned using computationally efficient shallow features, markedly accelerating the generation of proposed Regions of Interest (ROIs) and diminishing the computational demands for subsequent object detection and classification. Comparative analysis on various public and private datasets illustrates that DiffRank, while maintaining high accuracy, substantially outperforms existing baselines in terms of speed, particularly with larger image sizes (e.g., an improvement exceeding 300 % at 1920×1080 resolution). Moreover, the method demonstrates enhanced robustness compared to baseline methods, efficiently disregarding static targets like mannequins in display windows. The advancements in candidate area preprocessing enable a balanced approach between detection accuracy and overall detection speed, making the algorithm highly applicable for real-time on-site analysis in edge computing scenarios and cloud-edge collaborative computing environments.

Hazard Perception in Visually Impaired Drivers Who Use Bioptic Telescopes.

Bioptic telescopic spectacles can allow individuals with central vision impairment to obtain or maintain driving privileges. The purpose of this study was to (1) compare hazard perception ability among bioptic drivers and traditionally licensed controls, (2) assess the impact of bioptic telescopic spectacles on hazard perception in drivers with vision impairment, and (3) analyze the relationships among vision and hazard detection in bioptic drivers. Visual acuity, contrast sensitivity, and visual field were measured for each participant. All drivers completed the Driving Habits Questionnaire. Hazard perception testing was conducted using commercially available first-person video driving clips. Subjects signaled when they could first identify a traffic hazard requiring a change of speed or direction. Bioptic drivers were tested with and without their bioptic telescopes in alternating blocks. Hazard detection times for each clip were converted to z-scores, converted back to seconds using the average response time across all videos, and then compared among conditions. Twenty-one bioptic drivers and 21 normally sighted controls participated in the study. The hazard response time of bioptic drivers was improved when able to use the telescope (5.4 ± 1.4 seconds vs 6.3 ± 1.8 seconds without telescope); however, it remained significantly longer than for controls (4.0 ± 1.4 seconds). Poorer visual acuity, contrast sensitivity, and superior visual field sensitivity loss were related to longer hazard response times. Drivers with central vision loss had improved hazard response times with the use of bioptic telescopic spectacles, although their responses were still slower than normally sighted control drivers. The use of a bioptic telescope by licensed, visually impaired drivers improves their hazard detection speed on a video-based task, lending support to their use on the road.

Ensemble Random Forest-based Gradient Optimization based Energy Efficient Video Processing System for Smart Traffic Surveillance System

Deep learning solutions in big data applications can benefit cloud centres and can also lead to network communication overhead. Typically, data collected from traffic are sent to the traffic management centre for analysis. However, this process can worsen the network route to the traffic management centre. A two-tier mechanism has been developed to address this issue, which performs vehicle speed estimation and traffic congestion detection for efficient traffic management. The real-time traffic video data are captured and the video frames are initially processed through a foreground extraction process, which extracts the temporarily stopped vehicles on the road by removing background pixels from the frames. The video frames are then wrapped in an up-down view to remove the influence of the observation angle. The traffic congestion is then detected accurately based on the traffic characteristics using the proposed Ensemble Random Forest-based Gradient Optimization (ERF-GO) algorithm. The generalization error occurs when learning complex features on frames is minimized using a gradient-based optimization (GO) algorithm. Finally, the learned information on traffic conditions is forwarded to the cloud and edge computing environments based on network connection speed. The efficiency of the proposed ERF-GO is investigated in terms of performance metrics, namely root mean square error, speed detection error, execution time, computational cost, accuracy, latency, workload balance, precision, recall, f-measure, and congestion detection error rate. The analytic result displays that the proposed ERF-GO algorithm attains a greater accuracy rate of about 98.65% in detecting traffic congestion which is comparably higher than state-of-the-art methods.

YOLO-CCS: Vehicle detection algorithm based on coordinate attention mechanism

Accurate and fast vehicle detection is a key factor in the field of intelligent transportation. To solve the problems such as low recognition accuracy, slow detection speed, and poor robustness existing in current vehicle detection algorithms. This paper proposes a vehicle detection algorithm based on YOLOv5 and the coordinate attention mechanism (CA) named YOLO-CCS. YOLO-CCS enables the network to focus on the vehicle itself during the feature extraction process, reduces the loss of feature information and improves the effect of vehicle detection. In our algorithm, we firstly expand YOLOv5s with CA blocks at the location of feature extraction section of network backbone and enhance the extraction capabilities of key features, and suppress the interference of complex backgrounds by embedding position information. Secondly, to enable the network to extract rich feature information and improve the feature fusion capability of the network, we introduce the faster implementation of the Cross Stage Partial (CSP) Bottleneck with 2 convolutions (C2f) in the backbone and neck of the network, which has more residual blocks and skip connections, which provides rich feature semantic information, and heightens the feature extraction capabilities of the network. Thirdly, we integrate the SCYLLA-IOU (SIOU) loss function suitable for YOLO-CCS. So we can leverage the vector angle between the real box and the predicted box to further improve the accuracy of the algorithm and accelerate the convergence of the model. Experimental results show that, compared with the baseline model YOLOv5s, the mAP50 and mAP50−95 of our method increase by 3.2% and 1.7% respectively. Compared with other YOLO based models, mAP50 has increased by approximately 4%. And the detecting speed of YOLO-CCS algorithm reaches 48 FPS, which can meet the real-time requirements of vehicle detection.

Rapid identification of moldy peanuts based on three-dimensional hyperspectral object detection

Hyperspectral imaging (HSI) has been effectively used in the nondestructive assessment of food quality in recent years. However, the identification of moldy objects using HSIs still faces challenges, including slow detection speed and poor identification accuracy. To address these challenges, this study proposes a three-dimensional hyperspectral mold detection (3D-HMD) approach. The model utilizes multiple 3D convolution (3DMC) modules as the backbone network for optimizing spectral-spatial feature extraction and introduces an attention mechanism to promote the feature information of different hyperspectral bands. A feature pyramid network (FPN) is then used to fuse classification features outputted from the backbone network for feature enhancement. To improve the recognition efficiency for moldy targets, a detection head module derived from the field of object detection is introduced to achieve HIS object-level classification. The experimental results indicate that the detection speed of the proposed model is nearly tenfold greater than that of traditional algorithms, such as 1D-RNN and 3D-CNN, with a mean average precision (mAP) of 81.63 %. Overall, the 3D-HMD model demonstrates remarkable efficiency and accuracy in recognizing moldy peanuts, leading to suitable applications for food quality detection.