Search

Rock avalanche disasters in alpine and gorge regions are frequent and large in scale and cause severe damage. The movement of a rock avalanche is complex and has not been fully studied. The deposits of a rock avalanche can provide valuable insights into its movement process, which is crucial in understanding the rock fragmentation mechanism and predicting disaster-affected areas. Taking the Wangjiapo rock avalanche in Yunnan Province of China as an example, the size, shape and distribution characteristics of the deposit were analyzed based on field surveys, unmanned aerial vehicle (UAV) photography and image recognition technology. Initially, 3062 deposited rock blocks were manually measured in the field. Subsequently, the Particles/Pores and Cracks Analysis System (PCAS) was employed to identify 11,357 rock blocks with an area greater than 0.1 m2 from UAV orthophotos. By comparing the characteristics of the rock blocks obtained through image recognition and manual measurement, the statistical analysis of UAV aerial imagery combined with PACS proved feasible in studying the Wangjiapo rock avalanche. The results showed that the rock block movement was accompanied by fragmentation and sorting processes; furthermore, the roundness increased with the migration distance. Small blocks were more prevalent at the foot of the slope, while irregularly shaped, large blocks dominated in source areas. The movement of huge blocks was characterized by significant potential energy-driven features and inertia advantages, allowing them to travel farther than smaller blocks, and they tended to be concentrated in the central area of the deposit. Additionally, affected by the cementation degree of breccia and the topography, the blocks in the eastern and western deposit areas exhibited different fragmentation and deposition characteristics. [ABSTRACT FROM AUTHOR]

Field-road classification that automatically identifies the activity (either in-field or on-road) of each point in Global Navigation Satellite System (GNSS) trajectories is a critical process in the behavior analysis of agricultural vehicles. To capture movement patterns specific to agricultural operations, we propose a multi-view field-road classification method, which extracts a physical and a visual feature vector to represent a trajectory point. We propose a task-specific approach using a pre-trained visual model to effectively extract visual features. Firstly, an image is generated based on a point plus its neighboring points to provide the contextual information of the point. Then, an image recognition model, a fine-tuned ResNet model is developed using the pretraining-finetuning paradigm. In such a paradigm, a pre-training process is used to train an image recognition model (ResNet) with natural image datasets (e.g., ImageNet), and a fine-tuning process is applied to update the parameters of the pre-trained model using the trajectory point images, enabling the model to have both general knowledge and task-specific knowledge. Finally, a visual feature is extracted for a point by the fine-tuned model, thereby overcoming the limitations caused by the small-scale generated images. To validate the effectiveness of our multi-view field-road classification, we conducted experiments on four trajectory datasets (Wheat 2021, Paddy, Wheat 2023, and Wheat 2024). The results demonstrated that the proposed method achieves competitive accuracy performance, i.e., 92.56%, 87.91%, 90.31%, and 94.23% on four trajectory datasets, respectively. Extensive experiments demonstrate that our approach can consistently perform better than the existing state-of-the-art method on the four trajectory datasets by 2.99%, 4.42%, 2.88%, and 2.77% in the F1-score, respectively. In addition, we conduct an in-depth analysis to verify the necessity and effectiveness of our method. [ABSTRACT FROM AUTHOR]

As one of the hottest topics in medical imaging, artificial intelligence (AI) plays a major role in large-scale medical image processing. In many studies, it has been applied for automatic recognition of complex patterns in image data and different stages of treatment. As the world's population is aging, countries around the world face a rapid increase in the demand for elderly care. The use of AI in medical image recognition supports clinical staff in their image analysis and facilitates faster diagnosis when there is a shortage of caregivers. It is particularly useful for the treatment of chronic wounds. Given the complexity and variety of issues associated with acute and chronic wounds, they are typically documented in diagnostic and nursing records with text descriptions and photographs. However, wounds may remain inflamed for a long time owing to the nursing staff's poor understanding of wound care procedures, inadequate wound assessment skills, and insufficient knowledge of the use of dressings. Not only does this cause physical and psychological suffering to patients, but also results in high medical expenses. To this end, this study proposes an AIoT-based prognostic assessment system that supports the treatment of pressure injuries in residential facilities. The system uses an intelligent mobile IoT device to quickly measure wound size and changes in tissue ratios with a deep recognition module, which allows caregivers to quickly classify wound stages and estimate the healing condition based on the system's results. Clinically, it can help physicians and nursing staff to automatically generate a pressure injury report. This study conforms to the government's guidelines on wound assessment, which cover causes of wounds, treatment goals, and care procedures. The system is intended to support physicians in clinical decision-making and nursing staff in efficient reporting. In addition, the system can be used for tracking patients' health to reduce the risk of wound deterioration and to provide more targeted medical care for the elderly community. [ABSTRACT FROM AUTHOR]

Convolution neural networks have become a fundamental model for solving various computer vision tasks. However, these operations are only invariant to translations of objects and their performance suffer under rotation and other affine transformations. This work proposes a novel neural network that leverages geometric invariants, including curvature, higher-order differentials of curves extracted from object boundaries at multiple scales, and the relative orientations of edges. These features are invariant to affine transformation and can improve the robustness of shape recognition in neural networks. Our experiments on the smallNORB dataset with a 2-layer network operating over these geometric invariants outperforms a 3-layer convolutional network by 9.69% while being more robust to affine transformations, even when trained without any data augmentations. Notably, our network exhibits a mere 6% degradation in test accuracy when test images are rotated by 40 ∘ , in contrast to significant drops of 51.7 and 69% observed in VGG networks and convolution networks, respectively, under the same transformations. Additionally, our models show superior robustness than invariant feature descriptors such as the SIFT-based bag-of-words classifier, and its rotation invariant extension, the RIFT descriptor that suffer drops of 35 and 14.1% respectively, under similar image transformations. Our experimental results further show improved robustness against scale and shear transformations. Furthermore, the multi-scale extension of our geometric invariant network, that extracts curve differentials of higher orders, show enhanced robustness to scaling and shearing transformations. [ABSTRACT FROM AUTHOR]

Siamese networks excel at comparing two images, serving as an effective class verification technique for a single‐per‐class reference image. However, when multiple reference images are present, Siamese verification necessitates multiple comparisons and aggregation, often unpractical at inference. The Centre‐Loss approach, proposed in this research, solves a class verification task more efficiently, using a single forward‐pass during inference, than sample‐to‐sample approaches. Optimising a Centre‐Loss function learns class centres and minimises intra‐class distances in latent space. The authors compared verification accuracy using Centre‐Loss against aggregated Siamese when other hyperparameters (such as neural network backbone and distance type) are the same. Experiments were performed to contrast the ubiquitous Euclidean against other distance types to discover the optimum Centre‐Loss layer, its size, and Centre‐Loss weight. In optimal architecture, the Centre‐Loss layer is connected to the penultimate layer, calculates Euclidean distance, and its size depends on distance type. The Centre‐Loss method was validated on the Self‐Checkout products and Fruits 360 image datasets. Centre‐Loss comparable accuracy and lesser complexity make it a preferred approach over sample‐to‐sample for the class verification task, when the number of reference image per class is high and inference speed is a factor, such as in self‐checkouts. [ABSTRACT FROM AUTHOR]

Machine learning systems, particularly in the domain of image recognition, are susceptible to adversarial perturbations applied to input data. These perturbations, while imperceptible to humans, have the capacity to easily deceive deep learning classifiers. Current defense methods for image recognition focus on using diffusion models and their variants. Due to the depth of diffusion models and the large amount of computations generated during each inference process, the GPU and storage performance of the device are extremely high. To address this problem, we propose a new defense-based non-overlapping image compression filter for image recognition classifiers against adversarial attacks. This method inserts a non-overlapping image compression filter before the classifier to make the results of the classifier invariant under subtle changes in images. This method does not weaken the adversarial robustness of the model and can reduce the computational cost during the training process of the image classification model. In addition, our method can be easily integrated with existing image classification training frameworks with only some minor adjustments. We validate our results by performing a series of experiments under three different convolutional neural network architectures (VGG16, ResNet34, and Inception-ResNet-v2) and on different datasets (CIFAR10 and CIFAR100). The experimental results show that under the Inception-ResNet-v2 architecture, our method achieves an average accuracy of up to 81.15% on the CIFAR10 dataset, fully demonstrating its effectiveness in mitigating adversarial attacks. In addition, under the WRN-28-10 architecture, our method achieves not only 91.28% standard accuracy on the CIFAR10 dataset but also 76.46% average robust accuracy. The test experiment on the model training time consumption shows that our defense method has an advantage in time cost, proving that our defense method is a lightweight and efficient defense strategy. [ABSTRACT FROM AUTHOR]

Rapid economic growth and increasing urban population have led to a significant increase in waste production, raising serious concerns for countries worldwide. As the population expands, the increase in waste generation poses numerous environmental and public health challenges. This study focuses on educating children about recyclable waste to promote early awareness and proper waste classification habits. Specifically, this study investigates the performance of the YOLOv8 model to embed it into a waste recognition system tailored for children's waste management education. Datasets were obtained from Kaggle and underwent preprocessing. The findings show that a model with 100 epochs, an SGD optimizer, and a batch size of 25 achieved the best performance, with an accuracy of over 94 % and a low loss of 0.367. This model demonstrated competitive accuracy in detecting and classifying waste images, highlighting its potential as an effective tool in educational programs aimed at teaching children the importance of waste management and promoting sustainable practices from an early age. [ABSTRACT FROM AUTHOR]

Modernization of agriculture is associated with increased use of energy and inputs (fertilizers, plant protection products, water), which today need to be better managed for optimization purpose and to limit risks for people and the environment. This issue has given rise to smart agriculture, a relative new approach to farming based on the integration of information and communication technologies, particularly Artificial Intelligent (AI) and Internet of Things, in which the environment (soil, relief, air, etc.) represents potential data source that should be exploited. In this work, we propose a service of smart agriculture that based on plant image recognition. A robust and simple Convolutional Neural Network (CNN) model called Kamagate, Kopoin and Dagou Neural Network (KKDNet) is developed for plant image classification. For our first experiment, we focused on tomato cultivation, which is very demanding in terms of energy and inputs. The proposed model KKDNet with less layer than other CNN architecture achieves approximately the same performance in metric like accuracy. As concerning the execution time, the other models use more execution time. we need to multiply KKDNet's execution time by a coefficient ranging from 25 to 52 to match the time required for the other architectures discussed in this study. [ABSTRACT FROM AUTHOR]

In recent years, deep learning methods were good solutions for object detection in synthetic aperture radar (SAR) images. However, the problems of complex scenarios, large object scale differences and imperfect fine‐grained classification in ship detection were intractable. In response, an improved model GDB‐YOLOv5s (Improved YOLOv5s model incorporating global attention mechanism (GAM), DCN‐v2 and BiFusion) is designed. This model introduces deformable convolution networks (DCN‐v2) into the Backbone to enhance the adaptability of the receptive field. It replaces the original Neck's PANet structure with a BiFusion structure to better fuse the extracted multiscale features. Additionally, it integrates GAM into the network to reduce information loss and improve global feature interaction. Experiments were conducted on single‐class dataset SSDD and multi‐class dataset SRSSD‐V1.0. The results show that the GDB‐YOLOv5s model improves mean average precision scores (mAP) significantly, outperforming the original YOLOv5s model and other traditional methods. GDB‐YOLOv5s also improves Precision‐score (P) and Recall‐score (R) for fine‐grained classification to some extent, thereby reducing false alarms and missed detections. It has been proved that the improved model is relatively effective. [ABSTRACT FROM AUTHOR]

Introduction: In the field of facility agriculture, the accurate identification of tomatoes at multiple stages has become a significant area of research. However, accurately identifying and localizing tomatoes in complex environments is a formidable challenge. Complex working conditions can impair the performance of conventional detection techniques, underscoring the necessity for more robust methods. Methods: To address this issue, we propose a novel model of YOLOv8-EA for the localization and identification of tomato fruit. The model incorporates a number of significant enhancements. Firstly, the EfficientViT network replaces the original YOLOv8 backbone network, which has the effect of reducing the number of model parameters and improving the capability of the network to extract features. Secondly, some of the convolutions were integrated into the C2f module to create the C2f-Faster module, which facilitates the inference process of the model. Third, the bounding box loss function was modified to SIoU, thereby accelerating model convergence and enhancing detection accuracy. Lastly, the Auxiliary Detection Head (Aux-Head) module was incorporated to augment the network's learning capacity. Result: The accuracy, recall, and average precision of the YOLOv8-EA model on the self-constructed dataset were 91.4%, 88.7%, and 93.9%, respectively, with a detection speed of 163.33 frames/s. In comparison to the baseline YOLOv8n network, the model weight was increased by 2.07 MB, and the accuracy, recall, and average precision were enhanced by 10.9, 11.7, and 7.2 percentage points, respectively. The accuracy, recall, and average precision increased by 10.9, 11.7, and 7.2 percentage points, respectively, while the detection speed increased by 42.1%. The detection precision for unripe, semi-ripe, and ripe tomatoes was 97.1%, 91%, and 93.7%, respectively. On the public dataset, the accuracy, recall, and average precision of YOLOv8-EA are 91%, 89.2%, and 95.1%, respectively, and the detection speed is 1.8 ms, which is 4, 4.21, and 3.9 percentage points higher than the baseline YOLOv8n network. This represents an 18.2% improvement in detection speed, which demonstrates good generalization ability. Discussion: The reliability of YOLOv8-EA in identifying and locating multi-stage tomato fruits in complex environments demonstrates its efficacy in this regard and provides a technical foundation for the development of intelligent tomato picking devices. [ABSTRACT FROM AUTHOR]

Smart agriculture has become an inevitable trend in the development of modern agriculture, especially promoted by the continuous progress of large language models like chat generative pre-trained transformer (ChatGPT) and general language model (ChatGLM). Although these large models perform well in general knowledge learning, they still have certain limitations and errors when facing agricultural professional knowledge about crop disease identification, growth stage judgment, and so on. Agricultural data involves images and texts and other modalities, which play an important role in agricultural production and management. In order to better learn the characteristics of different modal data in agriculture, realize cross-modal data fusion, and thus understand complex application scenarios, we propose a framework AgriVLM that uses a large amount of agricultural data to fine-tune the visual language model to analyze agricultural data. It can fuse multimodal data and provide more comprehensive agricultural decision support. Specifically, it utilizes Q-former as a bridge between an image encoder and a language model to achieve a cross-modal fusion of agricultural images and text data. Then, we apply a Low-Rank adaptive to fine-tune the language model to achieve an alignment between agricultural image features and a pre-trained language model. The experimental results prove that AgriVLM demonstrates great performance in crop disease recognition and growth stage recognition, with recognition accuracy exceeding 90%, demonstrating its capability to analyze different modalities of agricultural data. [ABSTRACT FROM AUTHOR]

Background/Objective: This study aims to utilize advanced artificial intelligence (AI) image recog-nition technologies to establish a robust system for identifying features in lung computed tomog-raphy (CT) scans, thereby detecting respiratory infections such as SARS-CoV-2 pneumonia. Spe-cifically, the research focuses on developing a new model called Residual-Dense-Attention Gates U-Net (RDAG U-Net) to improve accuracy and efficiency in identification. Methods: This study employed Attention U-Net, Attention Res U-Net, and the newly developed RDAG U-Net model. RDAG U-Net extends the U-Net architecture by incorporating ResBlock and DenseBlock modules in the encoder to retain training parameters and reduce computation time. The training dataset in-cludes 3,520 CT scans from an open database, augmented to 10,560 samples through data en-hancement techniques. The research also focused on optimizing convolutional architectures, image preprocessing, interpolation methods, data management, and extensive fine-tuning of training parameters and neural network modules. Result: The RDAG U-Net model achieved an outstanding accuracy of 93.29% in identifying pulmonary lesions, with a 45% reduction in computation time compared to other models. The study demonstrated that RDAG U-Net performed stably during training and exhibited good generalization capability by evaluating loss values, model-predicted lesion annotations, and validation-epoch curves. Furthermore, using ITK-Snap to convert 2D pre-dictions into 3D lung and lesion segmentation models, the results delineated lesion contours, en-hancing interpretability. Conclusion: The RDAG U-Net model showed significant improvements in accuracy and efficiency in the analysis of CT images for SARS-CoV-2 pneumonia, achieving a 93.29% recognition accuracy and reducing computation time by 45% compared to other models. These results indicate the potential of the RDAG U-Net model in clinical applications, as it can accelerate the detection of pulmonary lesions and effectively enhance diagnostic accuracy. Additionally, the 2D and 3D visualization results allow physicians to understand lesions' morphology and distribution better, strengthening decision support capabilities and providing valuable medical diagnosis and treatment planning tools. [ABSTRACT FROM AUTHOR]

Asparagus stem blight, also known as "asparagus cancer", is a serious plant disease with a regional distribution. The widespread occurrence of the disease has had a negative impact on the yield and quality of asparagus and has become one of the main problems threatening asparagus production. To improve the ability to accurately identify and localize phenotypic lesions of stem blight in asparagus and to enhance the accuracy of the test, a YOLOv8-CBAM detection algorithm for asparagus stem blight based on YOLOv8 was proposed. The algorithm aims to achieve rapid detection of phenotypic images of asparagus stem blight and to provide effective assistance in the control of asparagus stem blight. To enhance the model's capacity to capture subtle lesion features, the Convolutional Block Attention Module (CBAM) is added after C2f in the head. Simultaneously, the original CIoU loss function in YOLOv8 was replaced with the Focal-EIoU loss function, ensuring that the updated loss function emphasizes higher-quality bounding boxes. The YOLOv8-CBAM algorithm can effectively detect asparagus stem blight phenotypic images with a mean average precision (mAP) of 95.51%, which is 0.22%, 14.99%, 1.77%, and 5.71% higher than the YOLOv5, YOLOv7, YOLOv8, and Mask R-CNN models, respectively. This greatly enhances the efficiency of asparagus growers in identifying asparagus stem blight, aids in improving the prevention and control of asparagus stem blight, and is crucial for the application of computer vision in agriculture. [ABSTRACT FROM AUTHOR]

Open-pit coal mining offers high resource recovery, excellent safety conditions, and large-scale production. However, the process generates significant dust, leading to occupational diseases such as pneumoconiosis among miners and adversely affecting nearby vegetation through dust deposition, which hinders photosynthesis and causes ecological damage. This limits the transition of open-pit mining to a green, low-carbon model. Among these processes, blasting generates the most dust and has the widest impact range, but the specific amount of dust generated has not yet been thoroughly studied. This study integrates indoor experiments, theoretical analyses, and field tests, employing the Split Hopkinson Pressure Bar (SHPB) system to conduct impact loading tests on coal–rock samples under pressures ranging from 0.13 MPa to 2.0 MPa. The results indicate that as the impact load increases, the proportion of large-sized blocks decreases while smaller fragments and powdered samples increase, signifying intensified sample fragmentation. Using stress wave attenuation theory, this study translates indoor impact loadings to field blast shock waves, revealing the relationship between blasting dust mass fraction and impact pressure. Field tests at the Haerwusu open-pit coal mine validated the formula. Using image recognition technology to analyze post-blast muck-pile fragmentation, the estimated dust production closely matched the calculated values, with an error margin of less than 10%. This formula provides valuable insights for estimating dust production and improving dust control measures during open-pit mine blasting operations. [ABSTRACT FROM AUTHOR]

The rapid and effective identification of leaf diseases of woody fruit plants can help fruit farmers prevent and cure diseases in time to improve fruit quality and minimize economic losses, which is of great significance to fruit planting. In recent years, deep learning has shown its unique advantages in image recognition. This paper proposes a new type of network based on deep learning image recognition method to recognize leaf diseases of woody fruit plants. The network merges the output of the convolutional layer of ResNet101 and VGG19 to improve the feature extraction ability of the entire model. It uses the transfer learning method to partially load the trained network weights, reducing model training parameters and training time. In addition, an attention mechanism is added to improve the efficiency of network information acquisition. Meanwhile, dropout, L2 regularization, and LN are used to prevent over-fitting, accelerate convergence, and improve the network's generalization ability. The experimental results show that the overall accuracy of woody fruit plant leaf diseases identification based on the model proposed in this paper is 86.41%. Compared with the classic ResNet101, the accuracy is improved by 1.71%, and the model parameters are reduced by 96.63%. Moreover, compared with the classic VGG19 network, the accuracy is improved by 2.08%, and the model parameters are reduced by 96.42%. After data set balancing, the overall identification accuracy of woody fruit plant leaf diseases based on the model proposed in this paper can reach 86.73%. [ABSTRACT FROM AUTHOR]

X‐ray flaw detection is widely used in non‐destructive area. The intuitive defect information can be obtained through the X‐ray film, which is usually digitized into high greyscale image by a 12‐bit or 16‐bit, called super 8‐bit, industrial scanner. When an ordinary 8‐bit monitor displays a super 8‐bit greyscale image, it appears loss of detail information, blurring of the image appears and other problems. Therefore, in this paper, a pseudo‐colour enhancement algorithm for displaying high‐bit RAW images on low‐bit monitor was proposed according to the chromatographic mapping relationship based on the visual characteristics of human eye. First, a high grey‐scale pseudo‐colour enhancement algorithm, called HOTM‐HGL, based on hot metal coding was proposed based on the standard film, whose enhancement effect is better than current mainstream algorithms. Second, aiming at the non‐standard film, a new stretching function called RAW‐Optical‐Stretching was reconstructed to improve HOTM‐HGL algorithm, called HOTM‐HGLS algorithm, whose display effect on ordinary monitors was improved in further. Finally, HOTM‐HGLS algorithm was applied in the detection of X‐ray film defect, which was convenient for the capture of defect information in the weld. Compared with the existing algorithms, various indicators have been greatly improved, enriching the amount of information and strengthening the image recognition effects. [ABSTRACT FROM AUTHOR]

The traditional robot vision system must keep the visual feature information in the process of visual servo stabilization, which greatly limits the environmental adaptability of mobile robots. In order to improve the effect of robot visual recognition, this paper combines information visualization algorithms to establish a set of sequential multi-free-surface spatial visual recognition methods, analyzes multiple visual recognition situations, and analyzes system performance in combination with optical algorithms. Moreover, this paper combines simulation experiment research to verify the system performance, constructs a robot vision system based on information visualization algorithm, and designs experiments to evaluate the robot vision system constructed in this paper. From the experimental research results, it can be seen that the new robot vision system constructed in this paper has a good visual recognition effect. [ABSTRACT FROM AUTHOR]

Face detection and key part extraction are the prerequisites for face recognition, and the goal is to obtain the current data set or image set face attribute features. This paper combines intelligent machine vision technology to construct a face recognition robot system. In order to solve the influence of noise environment and low resolution on AU recognition, this paper adopts a facial feature extraction method based on AU area. Moreover, in order to reduce the influence of noise factors on feature extraction, this paper preprocesses the image and combines image denoising and edge extraction methods to eliminate background. In addition, this paper constructs an intelligent image recognition algorithm based on the requirements of facial feature recognition, and uses this algorithm as the core algorithm of the robot. Finally, this paper designs experiments to verify the system. The experimental research results show that the face recognition robot system based on intelligent machine vision image recognition constructed in this paper has a good recognition effect. [ABSTRACT FROM AUTHOR]

In order to improve the operation effect of the railway train inspection robot, this paper applies image recognition technology to the robot system. The hardware part of the robot system includes: data acquisition station, data processing transfer station and inspection analysis center. First of all, this paper improves the traditional image recognition algorithm and builds an image recognition system suitable for railway train inspection requirements. Secondly, this paper combines the operating requirements of the railway train inspection robot to collect multiple sets of data through the vision system to establish a database, and collects trust data from the railway department to construct a standard database. In addition, this paper builds the intelligent identification system of this paper through simulation, obtains the railway train inspection robot, verifies the recognition performance of the railway train inspection robot through multiple sets of data, and counts the recognition accuracy rate. Finally, this paper verifies the reliability of the intelligent robot system constructed in this paper through experimental research. [ABSTRACT FROM AUTHOR]

With the advantages of low power consumption, flexibility, and high readability against bright ambiance, electrophoretic displays (EPDs) have wide application prospects in the fields of education, smart supermarkets, Internet of Things, smart homes, wearable devices, etc. EPDs suffer from a severe history dependence during grayscale switching, which results in annoying ghost images. However, currently, it is difficult to distinguish diverse types of ghost images automatically; thus, lookup-tables (LUTs) for multi-grayscale waveform design cannot be effectively generated but require cumbersome manual adjustment. In this article, we proposed to adopt a convolutional neural network (CNN) to automatically recognize ghost images, based on which, LUTs that could suppress ghost images and achieve accurate grayscales were automatically generated for waveform design. Moreover, the workforce for manual adjustment was significantly saved. The results suggest that the CNN is a powerful tool for EPDs to achieve better image quality, as well as less manual cost. [ABSTRACT FROM AUTHOR]

Acupressure is a noninvasive method in traditional Chinese medicine to increase blood circulation, relieve discomfort, promote health, and prevent disease. This paper presents a multifunctional acupoint application system that not only provides real-time visualization of acupoints on mobile devices for users' self-acupressure but also provides health care, education, and entertainment through acupressure. The system allows users to obtain acupoint information to relieve specific discomfort or symptoms. It automatically marks the requested acupoints on a real-time camera stream to guide users to locate acupoints so that even users unfamiliar with acupoints can easily find acupoints for self-acupressure. Three core technologies are used: acupoint localization and knowledge, real-time image recognition, and augmented reality. This paper details the system design and implementation. This system uses Google's machine learning kit for facial detection to extract the user's facial features and contours in real time on a mobile device. These contour points are used as landmarks with specific displacements to locate acupoints. As there are many acupoints and various positioning methods, the acupoint location is traditionally described in words through specific feature points and corresponding displacements related to each person's physiological characteristics. This work proposes a novel acupoint positioning and marking mechanism based on the standard acupoint localization, which formalizes various acupoint positioning methods so that positions can be processed and stored in an acupoint database established in this system. This positioning information allows the proposed mechanism to fulfill real-time visualization of acupoints in augmented reality on the camera stream. Experimental results show that the proposed system locates and marks acupoints with low latency and high accuracy. Moreover, the system also provides recommendations for further medical treatment and a mini-game that tests knowledge of acupoints, which enhances users' experience of acupoint massage. [ABSTRACT FROM AUTHOR]

Nowadays, physical dance is widely spread in the society as an emerging sport. Dance movement is favored by people because of its unique social function and fitness effect. For dance teaching, dance movement analysis can help optimize and improve the existing dance movements and the understanding and inheritance of traditional dance movements. With the rise of online teaching, intelligent identification and analysis of dance movements can promote the better development of sports dance teaching. However, the relevant research in this area is still very scarce. As the basis of this kind of research, there is an urgent need for dance motion recognition technology. Based on this background, this paper by introducing neural network algorithm for dance teaching sports image special recognition design, the algorithm can combine feature extraction technology to process video, extract the dance movements in the target data set, then for the extraction of cumulative feature extraction operation, in order to accumulate all the collected target features, so as to further complete the gradient histogram acquisition. Through the design experimental test, the cumulative feature image extraction results obtained through the algorithm are obviously better than the traditional image recognition results, so the design rationality and effectiveness of the algorithm are proved, and the sports dance teaching can be specially assisted. This paper designs an effective auxiliary image recognition algorithm by introducing the neural network algorithm into the field of sports dance teaching. [ABSTRACT FROM AUTHOR]

Under the new era background, the multi-cultures of various countries in the world are spreading and infiltrating each other, and the art education at this stage is in the stage of innovation and development. It is necessary to improve the teaching content and teaching methods of schools in an all-round way, deepen the research of education curriculum reform, especially to combine these new technologies with art classroom teaching to achieve the transformation of humanity. Today, with the great development trend of visual media, it has gradually entered the era of reading and map reading, which is a period of information flow and communication. The basic function of image is to transfer and convey image information. Therefore, in art teaching, it and graphics are complementary. In order to fully carry out art teaching, this paper applies the image recognition technology of machine vision feature parameters to art teaching and effect simulation. Machine vision technology analyzes the required resolution and color scale from digital imaging and uses preprocessed images and features to lock the images in the machine vision system, obtain the texture and proportion of art images according to the area, width and proportion of objects and complete classification and recognition. The whole image production process is more efficient, and finally can produce high-resolution art images. The results show that, due to the application of image recognition technology of machine vision feature parameters in art teaching and effect simulation, it has achieved good performance results. It has maximized the image recognition methods of students, thus cultivating students' artistic perception through artistic literacy, and promoting the balanced development of students. [ABSTRACT FROM AUTHOR]

Tibetan medicinal materials play a significant role in Tibetan culture. However, some types of Tibetan medicinal materials share similar shapes and colors, but possess different medicinal properties and functions. The incorrect use of such medicinal materials may lead to poisoning, delayed treatment, and potentially severe consequences for patients. Historically, the identification of ellipsoid-like herbaceous Tibetan medicinal materials has relied on manual identification methods, including observation, touching, tasting, and nasal smell, which heavily rely on the technicians' accumulated experience and are prone to errors. In this paper, we propose an image-recognition method for ellipsoid-like herbaceous Tibetan medicinal materials that combines texture feature extraction and a deep-learning network. We created an image dataset consisting of 3200 images of 18 types of ellipsoid-like Tibetan medicinal materials. Due to the complex background and high similarity in the shape and color of the ellipsoid-like herbaceous Tibetan medicinal materials in the images, we conducted a multi-feature fusion experiment on the shape, color, and texture features of these materials. To leverage the importance of texture features, we utilized an improved LBP (local binary pattern) algorithm to encode the texture features extracted by the Gabor algorithm. We inputted the final features into the DenseNet network to recognize the images of the ellipsoid-like herbaceous Tibetan medicinal materials. Our approach focuses on extracting important texture information while ignoring irrelevant information such as background clutter to eliminate interference and improve recognition performance. The experimental results show that our proposed method achieved a recognition accuracy of 93.67% on the original dataset and 95.11% on the augmented dataset. In conclusion, our proposed method could aid in the identification and authentication of ellipsoid-like herbaceous Tibetan medicinal materials, reducing errors and ensuring the safe use of Tibetan medicinal materials in healthcare. [ABSTRACT FROM AUTHOR]

The switching power supply is widely used in the secondary equipment of the power grid. The failure of the switching power supply will cause the protection device to malfunction, leading to a risk of large‐scale power failure. In this paper, a real‐time monitoring method based on only output voltage information is proposed. The experiment shows that this parameter‐reduced approach, relying only on voltage information, outperforms the existing method in detection accuracy. [ABSTRACT FROM AUTHOR]

The vision recognition module is one of the very important modules in the normal operation of the intelligent plant cultivation robot. In the original vision module, by improving the original YOLOV3(The algorithm proposed in YOLOv3: An Incremental Improvement), a more excellent YOLOV3 was obtained to recognize plants under normal conditions. However, the improved YOLOV3 does not perform well in the visual recognition process when there is a lot of noise interference. To make the visual recognition module continue to work properly, in this paper, the better Cycle GAN (generative adversarial networks) model is proposed to deal with noise based on three common noises, and finally, through a lot of experiments, it is proved that the improvement has some significance for the vision recognition module, which extends the use of the vision recognition module and makes the intelligent plant cultivation robot avoid the interference of noise. [ABSTRACT FROM AUTHOR]

Lightweight deep learning models for face recognition are becoming increasingly crucial for deployment on resource‐constrained devices such as embedded systems or mobile devices. This paper presents a highly efficient and compact deep learning (DL) model that achieves state‐of‐the‐art performance on various face recognition benchmarks. The developed DL model employs one‐ or few‐shot learning to obtain effective feature embeddings and draws inspiration from FaceNet with significant refinements to achieve a memory size of only 3.5 MB—about 30 times smaller than FaceNet—while maintaining high accuracy and real‐time performance. The study demonstrates the model's effectiveness through extensive experiments, which include testing on public datasets and the model's ability to recognize occluded faces in uncontrolled environments using grayscale input images. Compared to the state‐of‐the‐art lightweight models, the proposed model requires fewer FLOPs (0.06G), has a smaller number of parameters (1.2 M), and occupies a smaller model size (3.5 MB) while achieving a competitive level of recognition accuracy and real‐time performance. The results show that the model is well‐suited for deployment in embedded domains, including live entrance security checks, driver authorization, and in‐class attendance systems. The entire code of FN8 is available onGitHub. [ABSTRACT FROM AUTHOR]

This paper presents a novel dynamic camera parameter control method for the position and posture estimation of highly miniaturized AR markers (micro AR markers) using a low-cost general camera. The proposed method captures images from the camera at each cycle and detects markers from these images. Subsequently, it performs iterative calculations of the marker's position and posture to converge them to a specified accuracy while dynamically updating the camera's zoom, focus, and other parameter values based on the detected marker's depth distances. For a 10 mm square micro AR marker, the proposed system demonstrated recognition accuracy better than ± 1.0 % for depth distance and 2.5 ∘ for posture angle, with a maximum recognition range of 1.0 m. In addition, the iterative calculation time was 0.7 s for the initial detection of the marker. These experimental results indicate that the proposed method and system can be applied to the precise robotic handling of small objects at a low cost. [ABSTRACT FROM AUTHOR]

Neural architecture has been a research focus in recent years due to its importance in deciding the performance of deep networks. Representative ones include a residual network (ResNet) with skip connections and a dense network (DenseNet) with dense connections. However, a theoretical guidance for manual architecture design and neural architecture search (NAS) is still lacking. In this paper, we propose a manual architecture design framework, which is inspired by optimization algorithms. It is based on the conjecture that an optimization algorithm with a good convergence rate may imply a neural architecture with good performance. Concretely, we prove under certain conditions that forward propagation in a deep neural network is equivalent to the iterative optimization procedure of the gradient descent algorithm minimizing a cost function. Inspired by this correspondence, we derive neural architectures from fast optimization algorithms, including the heavy ball algorithm and Nesterov’s accelerated gradient descent algorithm. Surprisingly, we find that we can deem the ResNet and DenseNet as special cases of the optimization-inspired architectures. These architectures offer not only theoretical guidance, but also good performances in image recognition on multiple datasets, including CIFAR-10, CIFAR-100, and ImageNet. Moreover, we show that our method is also useful for NAS by offering a good initial search point or guiding the search space. [ABSTRACT FROM AUTHOR]

To measure the water level with a water level gauge, an automatic water level calculation method based on Region-based Convolutional Neural Networks (R-CNN) is proposed in this study. First, the water level gauge is located and the contour of water level gauge is obtained using the enhanced Mask R-CNN model approach. Secondly, using the contour of the water gauge, image processing technology is utilized to recognize and intercept the image that only contains the water gauge. Next, the scale on the water level gauge is detected and identified using the Faster R-CNN (Region-based Convolutional Neural Networks) model approach. The projection transformation method then calculates the value of the current water level based on the scale of the identified water level. The proposed approach was demonstrated by experimental findings to be relatively straightforward, to have high recognition accuracy and ability in complex interference environments in the field, and to have low cost and great application value. The findings presented in this paper can be used to help with automatic water level gauge measuring both theoretically and practically. [ABSTRACT FROM AUTHOR]

A complementary label specifies one of the classes that an instance does not belong to. Complementary label learning only uses training instances each assigned a complementary label to train a classifier that can predict a ground-truth label for each testing instance. Though many surrogate loss functions have been proposed for complementary label learning, the mean square error (MSE) surrogate loss function, widely used in the standard classification paradigm, cannot provide classifier consistency in complementary label learning. However, classifier consistency not only guarantees the converged model is the optimal classifier that can be found in the searching space but also indicates that standard backpropagation is enough to search for the optimal classifier without needing model selection. This paper designs an effective square loss for complementary label learning under unbiased and biased assumptions. We also theoretically demonstrate that our method assurances that the optimal classifier under complementary labels is also the optimal classifier under ordinary labels. Finally, we test our method on different benchmark datasets with biased and unbiased assumptions to verify the effectiveness of our method. [ABSTRACT FROM AUTHOR]

Copyright of Coal Science & Technology (0253-2336) is the property of Coal 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.)

Projective dictionary pair learning (DPL) and its variations have shown particular effectiveness for pattern classification due to their powerful discriminative capability and efficient computational performance. However, to boost up classification capacity, most of these methods mainly focus on adding more complicate discriminative terms, which not only sacrifices computational cost but also results in overfitting. To release the aforementioned issues, in this paper we propose a novel DPL-based classification framework which adopts an effective reconstruction strategy termed discriminative twin reconstructions (DTR) and a simple fisher-like discriminative term. Moreover, the proposed DTR strategy reconstructs data in a cross-data form with l 2 , 1 -norm constraint. The strategy benefits to simultaneously reduce the feature distribution gap of the coefficients in the same class and deliver more accurate reconstruction. But l 2 , 1 -norm sparse constraint is pivotal to achieve more efficient optimization yet discriminative capacity. The experimental results compared with 13 dictionary learning-based methods validate the impressive effectiveness of our proposed DTR-based method. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Data Acquisition & Processing / Shu Ju Cai Ji Yu Chu Li is the property of Editorial Department of Journal of Nanjing University of Aeronautics & Astronautics 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.)

Metal materials play a significant role in modern industrial fields. However, due to their exposure to harsh environments, they are susceptible to fatigue damage, which can lead to the occurrence of metal fractures (MF). Effective and accurate identification for MF types is crucial for preventing fractures during service and developing appropriate repair and maintenance strategies. In this paper, a multi-perception region of interest feature fusion (MRIFF) method based on deep learning is proposed to achieve accurate recognition of MF types. The proposed MRIFF method incorporates three key components – the mechanism exploration (ME) module, the adaptive voting (AVT) strategy, and the average-hybrid-attention (AHA) structure, where the ME module is designed to effectively extract and transfer features in metal fracture scanning electron microscope (MFSEM) images from three base models: VGG16, VGG19 and ResNet50, the AVT strategy is developed to synthesize information about MF extracted from these three base models by automatically adjust the weights of individual base model, and the AHA structure is designed to enhance the recognition ability for MF types by effectively fusing information from different scales. These components work synergistically to result in an efficient and accurate image recognition model for MF types. Experiments on the MFSEM image dataset and several publicly available datasets demonstrate that the proposed MRIFF method not only achieves higher accuracy when facing the recognition tasks of MFSEM image, but also exhibits better generalization performance on other image recognition tasks. [ABSTRACT FROM AUTHOR]

Insulators on transmission lines can be damaged to different degrees due to extreme weather conditions, which threaten the safe operation of the power system. In order to detect damaged insulators in time and meet the needs of real-time detection, this paper proposes a multi-defect and lightweight detection algorithm for insulators based on the improved YOLOv5s. To reduce the network parameters, we have integrated the Ghost module and introduced C3Ghost as a replacement for the backbone network. This enhancement enables a more efficient detection model. Moreover, we have added a new detection layer specifically designed for small objects, and embedded an attention mechanism into the network, significantly improving its detection capability for smaller insulators. Furthermore, we use the K-means++ algorithm to recluster the prior boxes and replace Efficient IoU Loss as the new loss function, which has better matching and convergence on the insulator defect dataset we constructed. The experimental results demonstrate the effectiveness of our proposed algorithm. Compared to the original algorithm, our model reduces the number of parameters by 41.1%, while achieving an mAP@0.5 of 94.8%. It also achieves a processing speed of 32.52 frames per second. These improvements make the algorithm well-suited for practical insulator detection and enable its deployment in edge devices. [ABSTRACT FROM AUTHOR]

Traditional manual defect detection methods were not efficient and accurate, so wood defect detection methods were gradually becoming intelligent mechanization. The application of digital image processing technology in wood defect recognition was widely promoted, and improving its accuracy, real-time performance, and robustness has always been the focus of exploration. In this paper, the application status of image processing in wood processing, including image preprocessing, image segmentation, image defect feature extraction and recognition was introduced, the principles, advantages and disadvantages of related algorithms were summarized, and the research direction of image processing in wood processing was prospected. [ABSTRACT FROM AUTHOR]

Convolutional Neural Network (CNN) is one of the most widely used deep learning models in pattern and image recognition. It can train a large number of datasets and get valuable results. The deep Residual Network (ResNet) is one of the most innovative CNN architecture to train thousands of layers or more and leads to high performance for complex problems. This deep model trains the neural networks with the idea of identity shortcut connection that skips layers. In this paper, we built a hyperparameter optimization approach for the ResNet models with different numbers of layers and show that our optimization leads to significant performance improvements. We developed a manual search approach by enhancing the traditional data augmentation proposed by the previous approaches. Using the CIFAR-100 dataset for classification, our approach significantly improves the classification error rate. It also reduces the computing time compared with the previous automatic approaches. Our approach leads to 23.01% error rate and 4 h and 15 min of computing time of ResNet-164 for the CIFAR-100 dataset using the NIVIDIA Tesla P100 GPU. [ABSTRACT FROM AUTHOR]

Crowdsourced online food images, when combined with food image recognition technologies, have the potential to offer a cost-effective and scalable solution for the assessment of the restaurant nutrition environment. While previous research has explored this approach and validated the accuracy of food image recognition technologies, much remains unknown about the validity of crowdsourced food images as the primary data source for large-scale assessments. In this paper, we collect data from multiple sources and comprehensively examine the validity of using crowdsourced food images for assessing the restaurant nutrition environment in the Greater Hartford region. Our results indicate that while crowdsourced food images are useful in terms of the initial assessment of restaurant nutrition quality and the identification of popular food items, they are subject to selection bias on multiple levels and do not fully represent the restaurant nutrition quality or customers' dietary behaviors. If employed, the food image data must be supplemented with alternative data sources, such as field surveys, store audits, and commercial data, to offer a more representative assessment of the restaurant nutrition environment. [ABSTRACT FROM AUTHOR]

How to quickly and accurately identify the bridge rubber bearing deterioration plays an important role in ensuring the bridge structure and road safety. This paper selects the common rubber bearings of domestic bridges as the research object, and proposes an improved YOLOv4-based bridge rubber bearing deterioration detection algorithm to address the reasons for the difficulty in detecting bridge rubber bearing deterioration due to large scale variations and small sample data sets. An image dataset (named HRBD) with annotations is constructed from real inspection scenarios, and the data is expanded by image processing means such as rotation, translation and brightness transformation, so that this dataset has sufficient data complexity and solves the problem of overfitting due to insufficient samples for network training. The anchor applicable to this dataset was regained by the K-means++ clustering algorithm, and then the CA module was inserted into the YOLOv4 backbone network for more accurate anchor localization. The improved YOLOv4 network was used for migration learning to train the dataset, and finally the trained network model was used for detection on the test set. The experimental results show that the improved YOLOv4 bridge rubber bearing deterioration detection and identification network can effectively identify and locate bridge rubber bearings and their deterioration types (crack damage, shear deformation, bearing void). [ABSTRACT FROM AUTHOR]

As the third-generation neural network, the spiking neural network (SNN) has become one of the most promising neuromorphic computing paradigms to mimic brain neural networks over the past decade. The SNN shows many advantages in performing classification and recognition tasks in the artificial intelligence field. In the SNN, the communication between the pre-synapse neuron (PRE) and the post-synapse neuron (POST) is conducted by the synapse. The corresponding synaptic weights are dependent on both the spiking patterns of the PRE and the POST, which are updated by spike-timing-dependent plasticity (STDP) rules. The emergence and growing maturity of spintronic devices present a new approach for constructing the SNN. In the paper, a novel SNN is proposed, in which both the synapse and the neuron are mimicked with the spin transfer torque magnetic tunnel junction (STT-MTJ) device. The synaptic weight is presented by the conductance of the MTJ device. The mapping of the probabilistic spiking nature of the neuron to the stochastic switching behavior of the MTJ with thermal noise is presented based on the stochastic Landau–Lifshitz–Gilbert (LLG) equation. In this way, a simplified SNN is mimicked with the MTJ device. The function of the mimicked SNN is verified by a handwritten digit recognition task based on the MINIST database. [ABSTRACT FROM AUTHOR]

Many Facial expression recognition methods have achieved great success, but they only considered front facial images or facial images close to the front. Besides, unlike in-the-laboratory datasets, the facial images in the real world (or in the wild) are without lighting and pose control, so that it is a big challenge to recognize these facial expressions. In this paper, the PIDViT (i.e., Pose-Invariant Distilled Vision Transformer) using the teacher-student model for the probability distributions of facial expressions of frontal and multi-pose faces was proposed and solved the pose variance and occlusion issues on expression recognition. First, the multi-pose face dataset FairFace-3D was built from the original FairFace and then used to train pose-invariance on the PIDViT. The PIDViT was trained in two stages; stage 1 is to train the PIDViT to achieve the consistency of facial expressions between frontal faces and multi-pose faces, and stage 2 is to use the student model pre-trained in stage 1 and train facial expressions further on target datasets. Finally, comprehensive experiments were conducted on three in the wild facial expression datasets, and the results validates the generalization of the PIDViT and its superiority over most state-of-the-art models. [ABSTRACT FROM AUTHOR]

As a safety-related application, visual systems based on deep neural networks (DNNs) in modern unmanned aerial vehicles (UAVs) show adversarial vulnerability when performing real-time inference. Recently, deep ensembles with various defensive strategies against adversarial samples have drawn much attention due to the increased diversity and reduced variance for their members. Aimed at the recognition task of remote sensing images (RSIs), this paper proposes to use a reactive-proactive ensemble defense framework to solve the security problem. In reactive defense, we fuse scoring functions of several classical detection algorithms with the hidden features and average output confidences from sub-models as a second fusion. In terms of proactive defense, we attempt two strategies, including enhancing the robustness of each sub-model and limiting the transferability among sub-models. In practical applications, the real-time RSIs are first input to the reactive defense part, which can detect and reject the adversarial RSIs. The accepted ones are then passed to robust recognition with a proactive defense. We conduct extensive experiments on three benchmark RSI datasets (i.e., UCM, AID, and FGSC-23). The experimental results show that the deep ensemble method of reactive and proactive defense performs very well in gradient-based attacks. The analysis of the applicable attack scenarios for each proactive ensemble defense is also helpful for this field. We also perform a case study with the whole framework in the black-box scenario, and the highest detection rate reaches 93.25%. Most of the adversarial RSIs can be rejected in advance or correctly recognized by the enhanced deep ensemble. This article is the first one to combine reactive and proactive defenses with a deep ensemble against adversarial attacks in the context of RSI recognition for DNN-based UAVs. [ABSTRACT FROM AUTHOR]

With the development of society and the continuous progress of science and technology, image recognition technology has developed rapidly. As a non-contact and non-destructive information monitoring technology, image recognition technology has been widely used in the field of chicken breeding, which has brought new impetus to the development of modern chicken industry. This paper briefly summarized the image recognition technology, then focused on the application of image recognition technology in chicken behavior recognition, body weight estimation, health monitoring and product classification, and finally summarized the problems and development prospects of image recognition technology in the field of chicken breeding, which was conducive to the efficient, sustainable and healthy development of image recognition technology in the future chicken industry, accelerated the comprehensive combination of artificial intelligence and chicken breeding industry, and provided technical reference basis for promoting the prosperity and development of modern animal husbandry in China. [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.)

Explainable AI (XAI) improved by a deep neural network (DNN) of a residual neural network (ResNet) and long short-term memory networks (LSTMs), termed XAIRL, is proposed for segmenting foot infrared imaging datasets. First, an infrared sensor imaging dataset is acquired by a foot infrared sensor imaging device and preprocessed. The infrared sensor image features are then defined and extracted with XAIRL being applied to segment the dataset. This paper compares and discusses our results with XAIRL. Evaluation indices are applied to perform various measurements for foot infrared image segmentation including accuracy, precision, recall, F1 score, intersection over union (IoU), Dice similarity coefficient, mean intersection of union, boundary displacement error (BDE), Hausdorff distance, and receiver operating characteristic (ROC). Compared to results from the literature, XAIRL shows the highest overall performance, achieving accuracy of 0.93, precision of 0.91, recall of 0.95, and F1 score of 0.93. XAIRL also displays the highest IoU, Dice similarity coefficient, and ROC curve and the lowest BDE and Hausdorff distance. Although U-Net performs well for most metrics, Mask R-CNN shows slightly worse performance but still outperforms the random forest and support vector machine algorithms. By building a high-quality foot infrared imaging dataset, machine learning-based algorithms can accurately analyze foot temperature and pressure distribution. These models can then be used to customize shoes for individual wearers, improving their comfort and reducing the risk of foot injuries, particularly for those with high blood pressure. [ABSTRACT FROM AUTHOR]

In order to effectively achieve the quality classification of sunflower seeds based on appearance identification in production scenarios, a multi-objective sunflower seed classification method based on a self-attention mechanism was proposed in this paper, which achieves the classification of sunflower seed object images with an accuracy of 94.71% by using a Multi-Head Self-attention mechanism to focus on the classification objects on the basis of effectively extracting context-dependent information from sunflower seed images. However, visualization experiments found that the model's attention could not effectively focus on the class activated regions of the image, resulting in a model that was heavily influenced by background noise and suffered from such defects as high computational cost and large model size. To address the above problems, a self-attention Focusing algorithm was proposed in this paper by evaluating and reconstructing the connections in the attention weight matrix, pruning the less influential connections, and realizing the update of the attention score, which gradually focuses attention on the class activated regions in continuous iterations, effectively shielding the background noise. Experiments have shown that applying the Focusing algorithm improves the accuracy rate to 96.6% based on a 34.3% reduction in the number of model parameters. Model accuracy was successfully improved while achieving model lightweighting, and the model was made more resistant to background noise due to the focused attention region. The self-attention Focusing algorithm's effectiveness in improving accuracy, focusing attention, and compressing model size is demonstrated. In practical production scenarios, it can be used as a popular application of key techniques for sunflower seed classification. [ABSTRACT FROM AUTHOR]

The multi-class classification of images is a pivotal challenge within the realm of image processing. As the volume of visual data continues to expand, there is a burgeoning interest in harnessing the unique capabilities of quantum computation to augment the efficiency of classification tasks. However, many existing methods for training quantum image multi-classifiers parallel classical machine learning techniques, where the requisite circuit measurements increase linearly with the volume of training data. This work introduces a novel approach for training a quantum image multi-classifier based on the quantum search algorithm. We have meticulously conducted rigorous experiments on a handwritten digit dataset, a classic benchmark in the field. The results have been meticulously compared with previous works, and the comparative analysis not only validates the efficiency of our proposed approach, requiring only O(N/b) measurements during training, but also highlights a significant quadratic speedup of the algorithm. [ABSTRACT FROM AUTHOR]

The multi-class classification of images is a pivotal challenge within the realm of image processing. As the volume of visual data continues to expand, there is a burgeoning interest in harnessing the unique capabilities of quantum computation to augment the efficiency of classification tasks. However, many existing methods for training quantum image multi-classifiers parallel classical machine learning techniques, where the requisite circuit measurements increase linearly with the volume of training data. This work introduces a novel approach for training a quantum image multi-classifier based on the quantum search algorithm. We have meticulously conducted rigorous experiments on a handwritten digit dataset, a classic benchmark in the field. The results have been meticulously compared with previous works, and the comparative analysis not only validates the efficiency of our proposed approach, requiring only O(N/b) measurements during training, but also highlights a significant quadratic speedup of the algorithm. [ABSTRACT FROM AUTHOR]