Target Feature Extraction Research Articles

Research on Marine Garbage Detection Based on Improved Yolov5 Model

The tidal movement of the ocean carries garbage to the shore. The garbage needs to be dealt with in time, otherwise, the pollution of the garbage to the environment will become increasingly serious. According to statistics, plastic garbage accounts for a substantial proportion of marine garbage. This study developed a target detection model for some plastic garbage to help achieve automatic marine garbage capture. Firstly, according to the principle of balanced label distribution, multi-background, and multi-angle, we created an image dataset based on artificial synthesis to solve the problem of insufficient data. Secondly, the CBAM attention module was used for the target detection algorithm Yolov5 to improve the ability of target feature extraction and model generalization. Furthermore, the loss function of bounding box regression CIoU was replaced with SIoU to solve the problems of slow convergence speed and low training efficiency. Finally, the effectiveness of the Yolov5 model was discussed with the analysis of experimental results.

Image-Free Target Classification With Semiactive Laser Detection System

Semiactive laser (SAL) guidance is a point-source detection technology without target feature extraction ability. In this article, the quadrant detector (QD)-based seeker and a spatially modulated designator are used in cooperation to achieve image-free target classification via a hybrid convolutional neural network (CNN) architecture. The convolutional layer is implemented optically by illuminating the target with training-obtained modulated patterns, and the detected intensity signals are regarded as compressed target features. The convolutional layers and classification net are trained together on an electronic platform. Layer normalization (LN) is adopted to improve network generalization ability and accelerate the training process. Network hyperparameters such as the channel numbers and kernel sizes of convolutional layers are analyzed and selected. The feasibility and robustness of the network are verified by both numerical simulation and experiments, whose classification accuracies are 88% and 66%, respectively. The proposed SAL detection and classification system is calculation-efficient and convenient to be adopted in existing guidance platforms or other real-time applications with limited computing resources.

An improved anchor-free SAR ship detection algorithm based on brain-inspired attention mechanism.

As a computing platform that can deal with problems independently and adapt to different environments, the brain-inspired function is similar to the human brain, which can effectively make use of visual targets and their surrounding background information to make more efficient and accurate decision results. Currently synthetic aperture radar (SAR) ship target detection has an important role in military and civilian fields, but there are still great challenges in SAR ship target detection due to the problems of large span of ship scales and obvious feature differences. Therefore, this paper proposes an improved anchor-free SAR ship detection algorithm based on brain-inspired attention mechanism, which efficiently focuses on target information ignoring the interference of complex background. First of all, most target detection algorithms are based on the anchor method, which requires a large number of anchors to be defined in advance and has poor generalization capability and performance to be improved in multi-scale ship detection, so this paper adopts an anchor-free detection network to directly enumerate potential target locations to enhance algorithm robustness and improve detection performance. Secondly, in order to improve the SAR ship target feature extraction capability, a dense connection module is proposed for the deep part of the network to promote more adequate deep feature fusion. A visual attention module is proposed for the shallow part of the network to focus on the salient features of the ship target in the local area for the input SAR images and suppress the interference of the surrounding background with similar scattering characteristics. In addition, because the SAR image coherent speckle noise is similar to the edge of the ship target, this paper proposes a novel width height prediction constraint to suppress the noise scattering power effect and improve the SAR ship localization accuracy. Moreover, to prove the effectiveness of this algorithm, experiments are conducted on the SAR ship detection dataset (SSDD) and high resolution SAR images dataset (HRSID). The experimental results show that the proposed algorithm achieves the best detection performance with metrics AP of 68.2% and 62.2% on SSDD and HRSID, respectively.

RMP-Net: A structural reparameterization and subpixel super-resolution-based marine scene segmentation network

Ocean exploration has always been an important strategic direction for the joint efforts of all mankind. Many countries in the world today are developing their own underwater autonomous explorers to better explore the seabed. Vision, as the core technology of autonomous underwater explorers, has a great impact on the efficiency of exploration. Different from traditional tasks, the lack of ambient light on the seabed makes the visual system more demanding. In addition, the complex terrain on the seabed and various creatures with different shapes and colors also make exploration tasks more difficult. In order to effectively solve the above problems, we combined the traditional models to modify the structure and proposed an algorithm for the super-resolution fusion of enhanced extraction features to perform semantic segmentation of seabed scenes. By using a structurally reparameterized backbone network to better extract target features in complex environments, and using subpixel super-resolution to combine multiscale feature semantic information, we can achieve superior ocean scene segmentation performance. In this study, multiclass segmentation and two-class segmentation tests were performed on the public datasets SUIM and DeepFish, respectively. The test results show that the mIoU and mPA indicators of our proposed method on SUIM reach 84.52% and 92.33%mPA, respectively. The mIoU and mPA on DeepFish reach 95.26% and 97.38%, respectively, and the proposed model achieves SOTA compared with state-of-the-art methods. The proposed model and code are exposed via Github1.

A Lightweight Model for Ship Detection and Recognition in Complex-Scene SAR Images

SAR ship detection and recognition are important components of the application of SAR data interpretation, allowing for the continuous, reliable, and efficient monitoring of maritime ship targets, in view of the present situation of SAR interpretation applications. On the one hand, because of the lack of high-quality datasets, most existing research on SAR ships is focused on target detection. Additionally, there have been few studies on integrated ship detection and recognition in complex SAR images. On the other hand, the development of deep learning technology promotes research on the SAR image intelligent interpretation algorithm to some extent. However, most existing algorithms only focus on target recognition performance and ignore the model’s size and computational efficiency. Aiming to solve the above problems, a lightweight model for ship detection and recognition in complex-scene SAR images is proposed in this paper. Firstly, in order to comprehensively improve the detection performance and deployment capability, this paper applies the YOLOv5-n lightweight model as the baseline algorithm. Secondly, we redesign and optimize the pyramid pooling structure to effectively enhance the target feature extraction efficiency and improve the algorithm’s operation speed. Meanwhile, to suppress the influence of complex background interference and ships’ distribution, we integrate different attention mechanism into the target feature extraction layer. In addition, to improve the detection and recognition performance of densely parallel ships, we optimize the structure of the model’s prediction layer by adding an angular classification module. Finally, we conducted extensive experiments on the newly released complex-scene SAR image ship detection and recognition dataset, named the SRSDDv1.0 dataset. The experimental results show that the minimum size of the model proposed in this paper is only 1.92 M parameters and 4.52 MB of model memory, which can achieve an excellent F1-Score performance of 61.26 and an FPS performance of 68.02 on the SRSDDv1.0 dataset.

Infrared dim and small target detection based on U-Transformer

Infrared dim and small target detection is a key technology for space-based infrared search and tracking systems. Traditional detection methods have a high false alarm rate and fail to handle complex background and high-noise scenarios. Also, the methods cannot effectively detect targets on a small scale. In this paper, a U-Transformer method is proposed, and a transformer is introduced into the infrared dim and small target detection. First, a U-shaped network is constructed. In the encoder part, the self-attention mechanism is used for infrared dim and small target feature extraction, which helps to solve the problems of losing dim and small target features of deep networks. Meanwhile, by using the encoding and decoding structure, infrared dim and small target features are filtered from the complex background while the shallow features and semantic information of the target are retained. Experiments show that anchor-free and transformer have great potential for infrared dim and small target detection. On the datasets with a complex background, our method outperforms the state-of-the-art detectors and meets the real-time requirement. The code is publicly available at https://github.com/Linaom1214/U-Transformer.

Solder Joint Defect Inspection Method Based on ConvNeXt-YOLOX

ConvNeXt-YOLOX model, an improved deep learning algorithm based on the YOLOX neural network, is proposed for surface mount technology (SMT) production line printed circuit board assembly (PCBA) solder joint defect detection with high accuracy and speed. The improved model employs the classification neural network ConvNeXt to enhance the feature extraction capability of the YOLOX backbone feature extraction network, thus enhancing the target detection accuracy and speed as well as the small target detection capability while maintaining the same number of parameters. A solder joint defect detection dataset containing 759 PCBA solder joint defect images is constructed, on which the improved model ConvNeXt-YOLOX, the original model YOLOX, and the lightweight model YOLOX-s are trained. Subsequently, the training results of the three models are compared. The comparison shows that the mean average precision (mAP) of the improved model ConvNeXt-YOLOX is 97.21%, which is 0.82% and 3.02% higher than that of YOLOX and YOLOX-s, respectively, while the mAP at (0.5:0.95) is increased from 76.3 to 77.5. Moreover, the detection speed is increased from the original 27.06 to 27.88 frames/s. In summary, the improved model ConvNeXt-YOLOX has strong small target feature extraction and detection capabilities, which are consistent with the actual requirements of solder joint defect detection.

Novel Cloud-Edge Collaborative Detection Technique for Detecting Defects in PV Components, Based on Transfer Learning

The existing techniques for detecting defects in photovoltaic (PV) components have some drawbacks, such as few samples, low detection accuracy, and poor real-time performance. This paper presents a cloud-edge collaborative technique for detecting the defects in PV components, based on transfer learning. The proposed cloud model is based on the YOLO v3-tiny algorithm. To increase the detection effect of small targets, we produced a third prediction layer by fusing the shallow feature information with the stitching layer in the second detection scale and introducing a residual module to achieve improvement of the YOLO v3-tiny algorithm. In order to further increase the ability of the network model to extract target features, the residual module was introduced in the YOLO v3-tiny backbone network to increase network depth and learning ability. Finally, through the model’s transfer learning and edge collaboration, the adaptability of the defect-detection algorithm to personalized applications and real-time defect detection was enhanced. The experimental results showed that the average accuracy and recall rates of the improved YOLO v3-tiny for detecting defects in PV components were 95.5% and 93.7%, respectively. The time-consumption of single panoramic image detection is 6.3 ms, whereas the consumption of the model’s memory is 64 MB. After cloud-edge learning migration, the training time for a local sample model was improved by 66%, and the accuracy reached 99.78%.

P‐2.4: Development and Prospect of Medical Action Recognition Technology forVirtual Reality (VR)

P‐2.4: Development and Prospect of Medical Action Recognition Technology forVirtual Reality (VR)

Comprehensive Biotransformation Analysis of Phenylalanine-Tyrosine Metabolism Reveals Alternative Routes of Metabolite Clearance in Nitisinone-Treated Alkaptonuria

Metabolomic analyses in alkaptonuria (AKU) have recently revealed alternative pathways in phenylalanine-tyrosine (phe-tyr) metabolism from biotransformation of homogentisic acid (HGA), the active molecule in this disease. The aim of this research was to study the phe-tyr metabolic pathway and whether the metabolites upstream of HGA, increased in nitisinone-treated patients, also undergo phase 1 and 2 biotransformation reactions. Metabolomic analyses were performed on serum and urine from patients partaking in the SONIA 2 phase 3 international randomised-controlled trial of nitisinone in AKU (EudraCT no. 2013-001633-41). Serum and urine samples were taken from the same patients at baseline (pre-nitisinone) then at 24 and 48 months on nitisinone treatment (patients N = 47 serum; 53 urine) or no treatment (patients N = 45 serum; 50 urine). Targeted feature extraction was performed to specifically mine data for the entire complement of theoretically predicted phase 1 and 2 biotransformation products derived from phenylalanine, tyrosine, 4-hydroxyphenylpyruvic acid and 4-hydroxyphenyllactic acid, in addition to phenylalanine-derived metabolites with known increases in phenylketonuria. In total, we observed 13 phase 1 and 2 biotransformation products from phenylalanine through to HGA. Each of these products were observed in urine and two were detected in serum. The derivatives of the metabolites upstream of HGA were markedly increased in urine of nitisinone-treated patients (fold change 1.2–16.2) and increases in 12 of these compounds were directly proportional to the degree of nitisinone-induced hypertyrosinaemia (correlation coefficient with serum tyrosine = 0.2–0.7). Increases in the urinary phenylalanine metabolites were also observed across consecutive visits in the treated group. Nitisinone treatment results in marked increases in a wider network of phe-tyr metabolites than shown before. This network comprises alternative biotransformation products from the major metabolites of this pathway, produced by reactions including hydration (phase 1) and bioconjugation (phase 2) of acetyl, methyl, acetylcysteine, glucuronide, glycine and sulfate groups. We propose that these alternative routes of phe-tyr metabolism, predominantly in urine, minimise tyrosinaemia as well as phenylalanaemia.

Two-Stage CNN Whole Heart Segmentation Combining Image Enhanced Attention Mechanism and Metric Classification.

Accurate segmentation of multiple tissues and organs in cardiac medical imaging is of great value in computer-aided cardiovascular diagnosis. However, it is challenging due to the complex distribution of various tissues and organs in cardiac MRI (magnetic resonance imaging) slices, low discriminative and large spanning organs. To handle these problems, a two-stage CNN (convolutional neural network) segmentation method based on the combination of Log-Gabor filter attention mechanism and metric classification is proposed. The Log-Gabor filterbank is applied to selectively enhance the texture information and contour information of each tissue and organ, and the spatial and channel attention mechanism jointly with the varying kernel size of Log-Gabor filterbank is incorporated into the codec structure to adaptively extract target features of different sizes and focus on the discriminative features in the network. To solve the problem of insufficient segmentation on subtle and adherent edges involving different tissues, a metric classification network is incorporated to finely optimize the hard-to-be-segmented boundaries. The proposed method was tested on cardiac MRI data set to segment 7 cardiac tissues, and the rationality and effectiveness of the method were verified. In comparison to a series of deep learning-based segmentation models, the proposed method achieves competitive performance.

DSA-SOLO: Double Split Attention SOLO for Side-Scan Sonar Target Segmentation

Side-scan sonar systems play an important role in tasks such as marine terrain exploration and underwater target identification. Target segmentation of side-scan sonar images is an effective method of underwater target detection. However, the principle of side-scan sonar systems leads to high noise interference, weak boundary information, and difficult target feature extraction of sonar images. To solve these problems, we propose a Double Split Attention (DSA) SOLO. Specially, we present an efficient attention module called DSA which fuses spatial attention and channel attention together effectively. DSA first splits feature maps into two parts along channel dimensions before processing them in parallel. Next, DSA utilizes C-S Unit and S-C Unit to describe relevant features in the spatial and channel dimensions, respectively. After that, the results of the two parts are aggregated to improve feature representation. We embedded the proposed DSA module after the FPN network of SOLOv2, and this approach improves the instance segmentation accuracy to a great extent. Experimental results show that our proposed DSA-SOLO on SCTD dataset achieves 78.4% mAP.5, which is 5.1% higher than SOLOv2.

DRUNet: A Method for Infrared Point Target Detection

Deep learning is widely used in vision tasks, but feature extraction of IR small targets is difficult due to the inconspicuous contours and lack of color information. This paper proposes a new convolutional neural network–based (CNN-based) method for IR small target detection called DRUNet. The algorithm is divided into two parts: the feature extraction network and the prediction head. For the small IR targets, which are difficult to accurately label, Gaussian soft labels are added to supervise the training process and make the network converge faster. We use a simplified object keypoint similarity to evaluate the network accuracy by the ratio of the distance to the radius of the inner tangent circle of the target box and a fair method for evaluating the model inference speed after GPU preheating. The experimental results show that our proposed algorithm performs better when compared with commonly used algorithms in the field of small target detection. The model size is 10.5 M, and the test speed reaches 133 FPS under the RTX3090 experimental platform.

Artificial Intelligence Technology to Record the Number of Times the Ball Passes the Net in Tennis Matches

Tennis competition is one of sports events. The vigorous development of sports can promote people’s pursuit of sports spirit and promote social unity and stability. However, in the course of social modernization development, sports events also need to follow closely, and we can no longer rely on traditional tools to support the development of events, which is not conducive to the positive development of sports. Due to the problem that the number of times the ball crosses the net in the traditional tennis game is too backward and the error rate is high, this paper uses artificial intelligence technology to record the number of times the ball crosses the net in tennis games and introduces the Center Net target based on the diagnostic criteria and comprehensive evaluation of tennis nets. Detect and track tennis balls by target recognition, feature extraction, and other methods, and record the times of net passing. However, considering that abnormal behaviors will also occur in the process of tennis passing through the net, the density-based DBSCAN clustering algorithm is used to discriminate and record abnormal behaviors in tennis matches. In order to verify the detection performance of DBSCAN clustering algorithm and Center Net target detection, the video of tennis match was analyzed and compared by drone from the perspective of time and space. By recording the number of times the ball crosses the net during the tennis receiving, connecting and attacking, and stalemate phases, the performance of the two recording times is compared. R 2 are all higher than 0.94, and R values are 0.982 and 0.963, respectively, so the recall rate of DBSCAN clustering algorithm is 0.02 higher, which is better than Center Net target detection. Using artificial intelligence technology to record the number of times the ball crosses the net in tennis games can not only improve the scientificity and accuracy of the record but also promote the development of the sports industry.

Research on the detection probability curve characteristics of long-range target based on SPAD array lidar

Using single photon avalanche diode (SPAD) array lidar to detect long-range target, the difference of target surface characteristics will cause the distortion of the detection probability curve. This distortion severely limits the ability of target feature extraction and recognition. In this paper, a detection probability model for detecting long-range target using SPAD array lidar is established, which considers the influence of the difference of target surface structural characteristics (distance and pose) and reflection characteristics (reflectivity and cross-sectional area). Based on the finite element analysis method, the proposed model can calculate the detection probability when the target surface has the above characteristics difference. The full width at half maximum (FWHM) and skewness are used to evaluate the distortion of the detection probability curve. It reveals that with the increase of the structural characteristic difference of the target, the increasing trend of FWHM satisfies the cubic polynomial, while the decreasing trend of skewness satisfies the cubic polynomial. As the ratio of the reflection characteristic of the target increases, the decreasing trend of FWHM satisfies the first-order polynomial, while the increasing trend of skewness satisfies the first-order polynomial. The experimental results show that the feature vector composed of the detection probability curve features (FWHM, skewness, and total detection probability) can be used to identify the pixels containing target surfaces with different distances. And the identifiable distance difference is less than the system pulse width distance. This paper lays a theoretical foundation for target characteristic analysis and target recognition based on SPAD array lidar.

Image Enhancement of Maritime Infrared Targets Based on Scene Discrimination.

Infrared image enhancement technology can effectively improve the image quality and enhance the saliency of the target and is a critical component in the marine target search and tracking system. However, the imaging quality of maritime infrared images is easily affected by weather and sea conditions and has low contrast defects and weak target contour information. At the same time, the target is disturbed by different intensities of sea clutter, so the characteristics of the target are also different, which cannot be processed by a single algorithm. Aiming at these problems, the relationship between the directional texture features of the target and the roughness of the sea surface is deeply analyzed. According to the texture roughness of the waves, the image scene is adaptively divided into calm sea surface and rough sea surface. At the same time, through the Gabor filter at a specific frequency and the gradient-based target feature extraction operator proposed in this paper, the clutter suppression and feature fusion strategies are set, and the target feature image of multi-scale fusion in two types of scenes are obtained, which is used as a guide image for guided filtering. The original image is decomposed into a target and a background layer to extract the target features and avoid image distortion. The blurred background around the target contour is extracted by Gaussian filtering based on the potential target region, and the edge blur caused by the heat conduction of the target is eliminated. Finally, an enhanced image is obtained by fusing the target and background layers with appropriate weights. The experimental results show that, compared with the current image enhancement method, the method proposed in this paper can improve the clarity and contrast of images, enhance the detectability of targets in distress, remove sea surface clutter while retaining the natural environment features in the background, and provide more information for target detection and continuous tracking in maritime search and rescue.

A Novel Efficient Convolutional Neural Algorithm for Multi-Category Aliasing Hardware Recognition.

When performing robotic automatic sorting and assembly operations of multi-category hardware, there are some problems with the existing convolutional neural network visual recognition algorithms, such as large computing power consumption, low recognition efficiency, and a high rate of missed detection and false detection. A novel efficient convolutional neural algorithm for multi-category aliasing hardware recognition is proposed in this paper. On the basis of SSD, the novel algorithm uses Resnet-50 instead of VGG16 as the backbone feature extraction network, and it integrates ECA-Net and Improved Spatial Attention Block (ISAB): two attention mechanisms to improve the ability of learning and extract target features. Then, we pass the weighted features to extra feature layers to build an improved SSD algorithm. At last, in order to compare the performance difference between the novel algorithm and the existing algorithms, three kinds of hardware with different sizes are chosen to constitute an aliasing scene that can simulate an industrial site, and some comparative experiments have been completed finally. The experimental results show that the novel algorithm has an mAP of 98.20% and FPS of 78, which are better than Faster R-CNN, YOLOv4, YOLOXs, EfficientDet-D1, and original SSD in terms of comprehensive performance. The novel algorithm proposed in this paper can improve the efficiency of robotic sorting and assembly of multi-category hardware.

Interrupted-Sampling Repeater Jamming-Suppression Method Based on a Multi-Stages Multi-Domains Joint Anti-Jamming Depth Network

Jamming will seriously affect the detection ability of radar, so it is essential to suppress the jamming of radar echoes. Interrupted-sampling repeater jamming (ISRJ) based on a digital-radio-frequency-memory (DRFM) device can generate false targets at the victim radar by the interception and repeating of the radar transmission signal, which is highly correlated with the true target signal. ISRJ can achieve main lobe jamming and has both deception and oppressive jamming effects, so it is difficult for the existing methods to suppress this jamming effectively. In this paper, we propose a deep-learning-based anti-jamming network, named MSMD-net (Multi-stage Multi-domain joint anti-jamming depth network), for suppressing ISRJ main lobe jamming in the radar echo. In the first stage of MSMD-net, considering that the target signal is difficult to detect under a high jamming-to-signal ratio (JSR), we propose a preprocessing method of limiting filtering on the time-frequency domain to reduce the JSR using the auxiliary knowledge of radar. In the second stage, taking advantage of the discontinuity of the jamming in the time domain, we propose a UT-net network that combines the U-net structure and the transformer module. The UT-net performs target feature extraction and signal reconstruction in the signal time-frequency domain and preliminarily realizes the suppression of the jamming component. In the third stage, combined with phase information, a one-dimensional complex residual convolution U-net network (ResCU-net) is constructed in the time domain to realize jamming filtering and signal recovery further. The experimental results show that MSMD-net can obtain the best jamming suppression effect under different transmitted signals, different jamming modes, and different jamming parameters.

Feature detection network-based correction method for accurate nano-tomography reconstruction.

Driven by the development of advanced x-ray optics such as Fresnel zone plates, nano-resolution full-field transmission x-ray microscopy (Nano-CT) has become a powerful technique for the non-destructive volumetric inspection of objects and has long been developed at different synchrotron radiation facilities. However, Nano-CT data are often associated with random sample jitter because of the drift or radial/axial error motion of the rotation stage during measurement. Without a proper sample jitter correction process prior to reconstruction, the use of Nano-CT in providing accurate 3D structure information for samples is almost impossible. In this paper, to realize accurate 3D reconstruction for Nano-CT, a correction method based on a feature detection neural network, which can automatically extract target features from a projective image and precisely correct sample jitter errors, is proposed, thereby resulting in high-quality nanoscale 3D reconstruction. Compared with other feature detection methods, even if the target feature is overlapped by other high-density materials or impurities, the proposed Nano-CT correction method still acquires sub-pixel accuracy in geometrical correction and is more suitable for Nano-CT reconstruction because of its universal and faster correction speed. The simulated and experimental datasets demonstrated the reliability and validity of the proposed Nano-CT correction method.

Remu-Net: Multi-Branch Net Framework for 3D Brain Tumor Segmentation

As one of the fatal human diseases, early detection of brain tumors can effectively save patients’ lives. Brain tumor image segmentation is of great practical importance for physicians to perform brain tumor diagnoses quickly. Due to the data complexity of 3D brain images, it is impractical to segment out tumor regions manually, so automatic and reliable methods can be utilized instead of manual work to achieve accurate segmentation of tumor regions. In this paper, we propose an end-to-end, more efficient brain tumor MRI segmentation model, REMU-Net, for the problems of multi-scale feature extraction and difficulty in small target feature extraction in 3D brain tumor image segmentation. Firstly, design and use the multi-channel parallel M-RepVGG module as a decoder to achieve multi-scale feature fusion. Secondly, embedding dilated convolution with different dilated rates in the DM-RepVGG module of the encoder to better extract features at different scales. Finally, introduce the expectation-maximizing attention in the network to better extract the features of the internal details of the tumor. The experimental results on the BraTS2018 validation dataset are Dice scores of 80.93%, 90.13%, and 86.15%, respectively. Experimental results on the BraTS2019 validation dataset can be achieved with Dice scores of 78.29%, 90.65%, and 82.77%, respectively.