Classical Detection Algorithms Research Articles

Advancing Ton-Bag Detection in Seaport Logistics with an Enhanced YOLOv8 Algorithm

Intelligent logistics and freight transportation is an important part of realizing the intelligence of port terminals. Due to the problems of inaccurate ton bag identification, high costs, large model sizes, and long computation times in traditional freight transportation—issues that hinder meeting real-time requirements on resource-constrained operational equipment—this paper proposes an improved lightweight ton bag detection algorithm, YOLOv8-TB (YOLOv8-Ton Bag), which is optimized based on YOLOv8. Firstly, the improved LZKAC module is introduced to combine with SPPF to form a new SPPFLKZ module, which improves the feature expression performance. Then, with reference to spatial and channel reconstruction convolution and deformable convolution, the C2f-SCTT block is designed for the backbone network, which reduces the spatial and channel redundancy between features in the network. Finally, the C2f-ORECZ block based on a linear scaling layer is designed for the neck, which reduces the training overhead and strengthens the feature learning of the feature extraction network for the targets in the complex background of the harbor and adds the 160 × 160 scale detection head to strengthen small target detection abilities. On the logistics ton bag operation dataset provided by shipping port enterprises, the improved algorithm improves by 3.7% and 5% compared with the original algorithm in mAP50 and mAP50-95, respectively, the model size is reduced by 4.42 MB and the amount of model computation is only 8 G, which is capable of accurately detecting logistics ton bags in real time. The superiority of the method is verified by comparing it with other classical target detection algorithms.

LES-YOLO: efficient object detection algorithm used on UAV for traffic monitoring

Abstract The use of UAVs for traffic monitoring greatly facilitates people’s lives. Classical object detection algorithms struggle to balance high speed and accuracy when processing UAV images on edge devices. To solve the problem, the paper introduces an efficient and slim YOLO with low computational overhead, named LES-YOLO. In order to enrich the feature representation of small and medium objects in UAV images, a redesigned backbone is introduced. And C2f combined with Coordinate Attention is used to focus on key features. In order to enrich cross-scale information and reduce feature loss during network transmission, a novel structure called EMS-PAN (Enhanced Multi-Scale PAN) is designed. At the same time, to alleviate the problem of class imbalance, Focal EIoU is used to optimize network loss calculation instead of CIoU. To minimize redundancy and ensure a slim architecture, the P5 layer has been eliminated from the model. And verification experiments show that LES-YOLO without P5 is more efficient and slimmer. LES-YOLO is trained and tested on the VisDrone2019 dataset. Compared with YOLOv8n-p2, mAP@0.5 and Recall has increased by 7.4% and 7%. The number of parameters is reduced by over 50%, from 2.9 M to 1.4 M, but there is a certain degree of increase in FLOPS, reaching 18.8 GFLOPS. However, the overall computational overhead is still small enough. Moreover, compared with YOLOv8s-p2, both the number of parameters and FLOPS are significantly reduced , while the performance is similar . As for real-time, LES-YOLO reaches 138 fps on GPU and a maximum of 78 fps on edge devices of UAV.

YOLO-ACT: an adaptive cross-layer integration method for apple leaf disease detection.

Apple is a significant economic crop in China, and leaf diseases represent a major challenge to its growth and yield. To enhance the efficiency of disease detection, this paper proposes an Adaptive Cross-layer Integration Method for apple leaf disease detection. This approach, built upon the YOLOv8s architecture, incorporates three novel modules specifically designed to improve detection accuracy and mitigate the impact of environmental factors. Furthermore, the proposed method addresses challenges arising from large feature discrepancies and similar disease characteristics, ultimately improving the model's overall detection performance. Experimental results show that the proposed method achieves a mean Average Precision (mAP) of 85.1% for apple leaf disease detection, outperforming the latest state-of-the-art YOLOv10s model by 2.2%. Compared to the baseline, the method yields a 2.8% increase in mAP, with improvements of 5.1%, 3.3%, and 2% in Average Precision, Recall, and mAP50-95, respectively. This method demonstrates superiority over other classic detection algorithms. Notably, the model exhibits optimal performance in detecting Alternaria leaf spot, frog eye leaf spot, gray spot, powdery mildew, and rust, achieving mAPs of 84.3%, 90.4%, 80.8%, 75.7%, and 92.0%, respectively. These results highlight the model's ability to significantly reduce false negatives and false positives, thereby enhancing both detection and localization of diseases. This research offers a new theoretical foundation and direction for future advancements in apple leaf disease detection.

Multiaxial low-cycle fatigue life model for notched specimens considering small sample characteristics

PurposeSufficient sample data are the necessary condition to ensure high reliability; however, there are relatively poor fatigue test data in the engineering, which affects fatigue life's prediction accuracy. Based on this, this research intends to analyze the fatigue data with small sample characteristics, and then realize the life assessment under different stress levels.Design/methodology/approachFirstly, the Bootstrap method and the principle of fatigue life percentile consistency are used to realize sample aggregation and information fusion. Secondly, the classical outlier detection algorithm (DBSCAN) is used to check the sample data. Then, based on the stress field intensity method, the influence of the non-uniform stress field near the notch root on the fatigue life is analyzed, and the calculation methods of the fatigue damage zone radius and the weighting function are revised. Finally, combined with Weibull distribution, a framework for assessing multiaxial low-cycle fatigue life has been developed.FindingsThe experimental data of Q355(D) material verified the model and compared it with the Yao’s stress field intensity method. The results show that the predictions of the model put forward in this research are all located within the double dispersion zone, with better prediction accuracies than the Yao’s stress field intensity method.Originality/valueAiming at the fatigue test data with small sample characteristics, this research has presented a new method of notch fatigue analysis based on the stress field intensity method, which is combined with the Weibull distribution to construct a low-cycle fatigue life analysis framework, to promote the development of multiaxial fatigue from experimental studies to practical engineering applications.

ODCS-YOLO detection algorithm for rail surface defects based on omni-dimensional dynamic convolution and context augmentation module

Abstract To solve the problems of easy miss and false detection on rail surface defects caused by small size, dense target, and high similarity between features and background, this paper proposed an improved detection algorithm in complex background. First, the conventional convolution of YOLOv5 backbone network is replaced with omni-dimensional dynamic convolution (ODConv), which improves the feature extraction capability of the network without increasing the computational cost; second, to improve the model's performance in detecting tiny objects, a two-layer context augmentation module(CAM) is introduced into the path aggregation network(PAN) structure; finally, the traditional non-maximum suppression(NMS) algorithm is replaced by the Soft-NMS algorithm in the network post-processing to reduce the false-alarm and miss-rate. The experimental results on the Railway Track Fault Detection public dataset show that the OD-YOLO (OD stands for ODConv) and C-PAN(CAM module is introduced into PAN) structures could achieve better performance in the same type of improved algorithms; compared with the baseline algorithm YOLOv5, the ODCS-YOLO (OD stands for ODConv, C stands for CAM, and S stands for Soft-NMS) algorithm improves the precision by 12.4%, the recall by 3.6%, the map50 by 8.6% and the GFLOPs is reduced by 0.6. Compared with seven classical object detection algorithms, the ODCS-YOLO algorithm achieves the highest detection accuracy, which makes it able to meet the real-time detection requirements of rail surface defects in real working conditions. The ODCS-YOLO model provides certain technical support for the defects detection and a new method for the detection of dense small objects.

An Infrared Aircraft Detection Algorithm Based on Context Perception Feature Enhancement

To address the issue of insufficient extraction of target features and the resulting impact on detection performance in long-range infrared aircraft target detection caused by small imaging area and weak radiation intensity starting from the idea of perceiving target context to enhance the features extracted by convolutional neural network, this paper proposes a detecting algorithm based on AWFGLC (adaptive weighted fusion of global–local context). Based on the mechanism of AWFGLC, the input feature map is randomly reorganized and partitioned along the channel dimension, resulting in two feature maps. One feature map is utilized by self-attention for global context modeling, establishing the correlation between target features and background features to highlight the salient features of the target, thereby enabling the detecting algorithm to better perceive the global features of the target. The other feature map is subjected to window partitioning, with max pooling and average pooling performed within each window to highlight the local features of the target. Subsequently, self-attention is applied to the pooled feature map for local context modeling, establishing the correlation between the target and its surrounding neighborhood, further enhancing the weaker parts of the target features, and enabling the detecting algorithm to better perceive the local features of the target. Based on the characteristics of the target, an adaptive weighted fusion strategy with learnable parameters is employed to aggregate the global context and local context feature maps. This results in a feature map containing more complete target information, enhancing the ability of the detection algorithm to distinguish between target and background. Finally, this paper integrates the mechanism of AWFGLC into YOLOv7 for the detection of infrared aircraft targets. The experiments indicate that the proposed algorithm achieves mAP50 scores of 97.8% and 88.7% on self-made and publicly available infrared aircraft datasets, respectively. Moreover, the mAP50:95 scores reach 65.7% and 61.2%, respectively. These results outperform those of classical target detection algorithms, indicating the effective realization of infrared aircraft target detection.

YOLOX-CS: An Automatic Search Algorithm for Low Surface Brightness Galaxies

The characteristics of Low Surface Brightness Galaxies (LSBGs) are very important for understanding the overall characteristics of galaxies. It is of great significance to search and expand the samples of Low Surface Brightness Galaxies by modern machine learning, especially deep learning algorithm. LSBGs are difficult to discern automatically and accurately with traditional methods because of their obscure features. However, deep learning does have the advantage of automatically identifying complex and effective features. To solve this problem, an algorithm named You Only Look Once version X-CS (YOLOX-CS) is proposed to search LSBG in large sample sky survey. Firstly, five classical target detection algorithms are compared through experiments and the optimal YOLOX algorithm is selected as the basic algorithm. Then, the YOLOX-CS framework is constructed by combining different attention mechanisms and different optimizers. The data set uses images from the Sloan Digital Sky Survey (SDSS), labelled from LSBG in the α.40-SDSS DR7 (the cross coverage area of 40% HI Arecibo Legacy Fast ALFA Survey and SDSS Data Release7) survey. Due to the small number of samples in this data set, Deep Convolutional Generative Adversarial Networks (DCGAN) model is used to expand the experimental test data. After comparing with a series of target detection algorithms, YOLOX-CS has a good test result in searching LSBG recall rate and Average Precision (AP) value in two data sets before and after expansion. The recall rate and AP value in the test set without expansion data set reach 97.75% and 97.83%, respectively. In the expanded data set of DCGAN model, under the same test set, the recall rate reaches 99.10% and the AP value reaches 98.94%, which proves that the algorithm has excellent performance in LSBG search. Finally, the algorithm is applied to SDSS photometric data, and 765 LSBG candidates are obtained.

YOLO-Chili: An Efficient Lightweight Network Model for Localization of Pepper Picking in Complex Environments

Currently, few deep models are applied to pepper-picking detection, and existing generalized neural networks face issues such as large model parameters, prolonged training times, and low accuracy. To address these challenges, this paper proposes the YOLO-chili target detection algorithm for chili pepper detection. Initially, the classical target detection algorithm YOLOv5 serves as the benchmark model. We introduce an adaptive spatial feature pyramid structure that combines the attention mechanism and the concept of multi-scale prediction to enhance the model’s detection capabilities for occluded and small target peppers. Subsequently, we incorporate a three-channel attention mechanism module to improve the algorithm’s long-distance recognition ability and reduce interference from redundant objects. Finally, we employ a quantized pruning method to reduce model parameters and achieve lightweight processing. Applying this method to our custom chili pepper dataset, we achieve an average precision (AP) value of 93.11% for chili pepper detection, with an accuracy rate of 93.51% and a recall rate of 92.55%. The experimental results demonstrate that YOLO-chili enables accurate and real-time pepper detection in complex orchard environments.

Human behaviour detection dataset (HBDset) using computer vision for evacuation safety and emergency management

During emergency evacuation, it is crucial to accurately detect and classify different groups of evacuees based on their behaviours using computer vision. Traditional object detection models trained on standard image databases often fail to recognise individuals in specific groups such as the elderly, disabled individuals and pregnant women, who require additional assistance during emergencies. To address this limitation, this study proposes a novel image dataset called the Human Behaviour Detection Dataset (HBDset), specifically collected and annotated for public safety and emergency response purposes. This dataset contains eight types of human behaviour categories, i.e. the normal adult, child, holding a crutch, holding a baby, using a wheelchair, pregnant woman, lugging luggage and using a mobile phone. The dataset comprises more than 1,500 images collected from various public scenarios, with more than 2,900 bounding box annotations. The images were carefully selected, cleaned and subsequently manually annotated using the LabelImg tool. To demonstrate the effectiveness of the dataset, classical object detection algorithms were trained and tested based on the HBDset, and the average detection accuracy exceeds 90 %, highlighting the robustness and universality of the dataset. The developed open HBDset has the potential to enhance public safety, provide early disaster warnings and prioritise the needs of vulnerable individuals during emergency evacuation.

Perception simplex: Verifiable collision avoidance in autonomous vehicles amidst obstacle detection faults

AbstractAdvances in deep learning have revolutionized cyber‐physical applications, including the development of autonomous vehicles. However, real‐world collisions involving autonomous control of vehicles have raised significant safety concerns regarding the use of deep neural networks (DNNs) in safety‐critical tasks, particularly perception. The inherent unverifiability of DNNs poses a key challenge in ensuring their safe and reliable operation. In this work, we propose perception simplex ( ), a fault‐tolerant application architecture designed for obstacle detection and collision avoidance. We analyse an existing LiDAR‐based classical obstacle detection algorithm to establish strict bounds on its capabilities and limitations. Such analysis and verification have not been possible for deep learning‐based perception systems yet. By employing verifiable obstacle detection algorithms, identifies obstacle existence detection faults in the output of unverifiable DNN‐based object detectors. When faults with potential collision risks are detected, appropriate corrective actions are initiated. Through extensive analysis and software‐in‐the‐loop simulations, we demonstrate that provides deterministic fault tolerance against obstacle existence detection faults, establishing a robust safety guarantee.

Vehicle Detection Algorithms for Autonomous Driving: A Review.

Autonomous driving, as a pivotal technology in modern transportation, is progressively transforming the modalities of human mobility. In this domain, vehicle detection is a significant research direction that involves the intersection of multiple disciplines, including sensor technology and computer vision. In recent years, many excellent vehicle detection methods have been reported, but few studies have focused on summarizing and analyzing these algorithms. This work provides a comprehensive review of existing vehicle detection algorithms and discusses their practical applications in the field of autonomous driving. First, we provide a brief description of the tasks, evaluation metrics, and datasets for vehicle detection. Second, more than 200 classical and latest vehicle detection algorithms are summarized in detail, including those based on machine vision, LiDAR, millimeter-wave radar, and sensor fusion. Finally, this article discusses the strengths and limitations of different algorithms and sensors, and proposes future trends.

231 Anomaly detection for breed purity analysis

Abstract Recent advancements in genotyping technologies have revolutionized our ability to estimate breed composition in pigs. The classic compositional regression used for this purpose requires a multibreed panel to detect crossbreeding and its application is limited to breed in the panel. In this study, we presented a protocol based on classic anomaly detection algorithms that use single breed panels to detect crossbreeding. By utilizing these methods, we aim to identify and assess breed outliers within large pig genotype panels. The dataset consisted of 44,616 SNP genotypes recorded in purebred Large White, Landrace, Duroc, Pietrain (n = 2,000 from each breed) and 1,010 crossbred pigs (50% Pietrain, 25% Large White, 25% Landrace). Large White pigs (n = 1,500) were randomly selected as reference set. The remaining 500 Large White pigs, other purebred pigs, crossbred pigs, and simulated crossbred animals of known proportion of Large White segments were test sets. We applied two anomaly detection methods: 1) Principal Component Analysis (PCA) and 2) deep learning autoencoders (AE). For PCA the top 600 eigen vectors capturing 85% cumulative proportion of variance in the reference set were selected. Test set genotypes were projected onto the principal components and then they were reconstructed to the full genotype set. For AE, we used a previously published sparse convolutional denoising autoencoder (SCDA) that consisted of two main components: an encoder and a decoder. The encoder compressed the genotypes to 128 components and the decoder reconstructed the set to the original 44,616 SNP. We utilized the model trained in the reference set to reconstruct test set genotypes of the same breed, of alternative breed and of crossbred. Accuracy of reconstruction of genotypes was computed using correlation and mean square error. The proportion of rejected genotypes was used to assess the ability of anomaly detection algorithms to detect crossbreeding. For each criterion, the threshold to reject genotypes was selected as the value at which 5% of Large White animals were rejected (specificity = 0.95). Results are summarized in (Table 1). PCA provided better properties to detect genomic outliers compared with AE. PCA anomaly detection was able to detect purebred individuals with probability 1.0. It also detected over 90% of crossbred individuals that were 75% Large Whites. Crossbred with a lower proportion of Large White genomic segments were detected with P < 0.5. Anomaly detection algorithms can be used to detect genomic outliers but with relatively low sensitivity. More work is needed to optimize these algorithms for breeding applications.

Damage Object Detection of Steel Wire Rope-Core Conveyor Belts Based on the Improved YOLOv5

In response to the challenges in detecting damage features in X-ray images of steel wire rope-cores in conveyor belts, such as complex damage shapes, small sizes, low detection precision, and poor generalization ability, an improved YOLOv5 algorithm was proposed. The aim of the model is to accurately and efficiently identify and locate damage in the X-ray images of steel wire rope-cores in conveyor belts. First, the Adaptive Histogram Equalization (AHE) method is used to preprocess the images, reducing the interference of harsh mining environments and improving the quality of the dataset. Second, to better retain image details and enhance the detection ability of damage features, transpose convolutional upsampling is adopted, and the C3 module in the backbone network is replaced by C2f to ensure lightweight network models, meanwhile, it obtains richer gradient flow information and optimizing the loss function. Finally, the improved algorithm is compared with four classical detection algorithms using the damage feature dataset of steel wire rope-core conveyor belts. The experimental result shows that the proposed algorithm achieves an average detection precision of 91.8% and a detection speed of 40 frames per second (FPS) for images collected in harsh mining environments. The designed detection model provides a reference for the automatic recognition and detection of damage to steel wire rope-core conveyor belts.

A comparison of deep learning-based object detection for unmanned aerial vehicle

Unmanned aerial vehicle (UAV) plays a critical role in the field of object detection, and machine learning techniques have significantly advanced the field in recent years. This paper provides a comprehensive overview of machine learning developments for UAV object detection. The process involves multiple steps. Firstly, deep learning and Convolutional Neural Network (CNN) are widely utilized to extract precise features and train object detection models for accurate classification and efficient recognition of target objects in images and videos. Secondly, classical object detection algorithms such as Faster R-CNN, You Only Look Once (YOLO), and Single Shot MultiBox Detector (SSD), have been enhanced to improve accuracy and real-time performance. In this review, we primarily focus on comparing the principles of CNN and YOLOv5 themselves, as well as their applications in object detection and image recognition. Ultimately, it becomes apparent that CNN is better suited for processing image data, automatically extracting features, and achieving more accurate classification and detection. On the other hand, YOLOv5 directly performs detection on large images, significantly reducing computation time compared to CNN.

Estimasi Arah Robot Pembersih Panel surya Menggunakan Algoritma Line Detection Klasik

Currently, solar panels are widely used in homes, offices, and buildings. Solar energy is converted into electrical energy by solar panels. The surface of solar panels determines how efficient they are; if the surface is dirty, the efficiency of solar panels will decrease. Therefore, they must be cleaned to maintain their effectiveness. Cleaning solar panels is one of the tasks that can be performed by an autonomous robot, which can move without the need for human control or remote control usage. To make the robot move autonomously, a camera is used to capture images of solar panels so that the robot can move on its own. The images captured by the camera will then be processed using a classic line detection algorithm (Hough line transform) to estimate the direction, in the form of the angle (theta), of the robot relative to the solar panel based on the lines detected by the algorithm. The direction of the robot relative to the solar panel determines the robot's future movement. The algorithm used (Hough line transform) has been able to estimate the direction of the robot quite well with a relatively fast processing time. Thus, it can be concluded that this research successfully determines the direction based on the available images

Deep Learning-Based Lane Detection: An In-depth Analysis

Lane detection technology is mainly studied in the traditional method at first, but this method is not suitable for practice. Because in some unfavorable conditions usually do not have the required robustness of autonomous driving systems. Later, Deep learning-based techniques that could increase recognition speed and accuracy were put forth. However, in the complex and changing environment, the recognition accuracy would be reduced, so the current methods need to be strengthened. This paper compares the classical methods based on image segmentation and object detection. The classical object detection algorithms such as VPGNet, Spatial CNN and spatial-temporal deep learning are introduced. In addition, the classical image segmentation methods such as Lanenet, UNet-ConvLSTM and MMA-Net are also introduced. By comparing the performance of these methods on different data sets, the advantages and problems of these methods are analyzed, and the development trend of these methods in this field is predicted. Overall, at the end of this review, the full text is summarized and prospected.

Advancing Hematological Point-of-Care Diagnostics: Development and Multi-Site Evaluation of a Novel AI-Enabled Diagnostic Device for CBC Analysis with Differential

Introduction: A complete blood count (CBC) with differential is a routine blood test that provides essential health information that is fundamental for maintaining wellness. While many other routine tests can be run in Point of Care (POC) settings, a CBC with differential is largely confined to central laboratory testing environments. Truvian is in late-stage development of a fully automated device that runs a TruWellness Panel (TM) incorporating all routine tests - a CBC with 3-part differential, comprehensive metabolic panel (CMP), lipid panel, HbA1c and TSH in a single run with 300uL of whole blood. This device incorporates high-throughput cell imaging, Computer Vision (CV) and Artificial Intelligence (AI), including machine learning and deep learning, for on board CBC analysis with 3-part and 5-part differentials. Here, we summarize our development and evaluation of the AI-enabled CBC sub-panel through method comparison studies. Methods: An AI-Enabled 5-part white blood cell (WBC) differential was developed using transfer learning. WBC cells were detected through classical object detection algorithms and extracted from fluorescent images of normal donor blood samples acquired on the Truvian device. Each cell in the set was classified by hematopathologists into one of five categories - Neutrophil, Lymphocyte, Monocyte, Eosinophil, and Basophil - to train a deep learning model. The algorithm currently outputs the data in a 3-part differential format (neutrophils, lymphocytes and “others”) with refinement of the full 5-part differential in progress. The final training dataset consisted of ~2000 - 5000 cells in each class. The training set was divided into training (80%) and test (20%) datasets. Data augmentation techniques were leveraged to improve model invariance to cell orientation and reduce the risk of overfitting. Performance of each iteration of the algorithm was assessed using accuracy of the model on test data. Following incorporation of the algorithm into Truvian's comprehensive and automated blood testing platform, comparability between an external reference laboratory and Truvian's system was performed. Samples from over 150 donors were measured and analytical concordance was evaluated by Passing-Bablok regression, and bias was evaluated using the TOST equivalency test. Results: WBC differential models were improved over a two-year period to incorporate cell image refinement associated with hardware and software iterations. The final algorithm was able to accurately detect and differentiate WBCs into 3 main categories: neutrophils, lymphocytes, and “others” (monocytes, eosinophils, and basophils). Neutrophils, lymphocytes, and “others” demonstrated R correlations of 0.98, 0.96 and 0.49 respectively. Slopes of 1.0 were observed for both neutrophils and lymphocytes while “others” had a slope of 1.37. Intercepts ranged from -0.07 to 0.04 x10 3 cells/μL for the various cell classes. The RBC, WBC and platelet cell counting algorithms were able to produce excellent correlation compared to predicate devices with correlation coefficients ≥ 0.95 and slopes between 0.96 to 1.03. Conclusions: In conclusion, this novel diagnostic device represents a breakthrough in point of care diagnostics. The compact, fully automated blood diagnostic device can run a CMP, lipid panel, CBC with 3-part differential, HbA1c and TSH assays simultaneously. By incorporating Computer Vision algorithms for cell detection and AI techniques such as machine and deep learning for differentiation on board the instrument, it offers an efficient tool for accurate and precise CBC analysis with 3-part differential at the point-of-action. Further development and refinement of the differential algorithm is currently underway.

A dilated convolution-based feature adaptation method for detection of high aspect ratio objects in aerial images

In real scenarios, objects with high aspect ratios are actually very common, and such objects hold significant importance in the field of object detection. However, most of the existing object detection algorithms tend to overlook this specific type of object. After analyzing the statistical data, we observed a substantial decrease in mAP (mean Average Precision) for classical object detection algorithms when they are tasked with detecting only high aspect ratio objects. Therefore, we conducted an analysis of the factors that influence the detection performance of these objects and made the following improvements: (1) We introduced large-kernel attention convolution between the backbone network layers. This addition allows each position feature to have a larger receptive field, facilitating better feature learning; (2) By incorporating multiple sets of deformable convolutions for feature-adaptive processing, we were able to enhance the learning of characteristic information specific to the object itself. This approach also promotes network convergence. The proposed method yielded a significant improvement in accuracy, approximately 5[Formula: see text] higher than the baseline, when evaluated on the FGSD2021 dataset. Furthermore, our method outperformed the current best method by approximately 0.5[Formula: see text].

DSW-YOLO: A detection method for ground-planted strawberry fruits under different occlusion levels

The problems of stem and leaf shading, fruit overlapping, and different ripening stages pose a great challenge to picking robots, which are usual for the ground-planted strawberries in field. To achieve accurate detection of ripe strawberries and their occlusion levels, a DSW-YOLO (DCNv3-SA-WIoU-YOLO) network model is proposed based on the YOLOv7 network model. The model is designed with a DCN-ELAN module by introducing DCNv3 (Deformable Convolution v3) in the ELAN module to improve the feature extraction capability of the network for non-regular targets. Meanwhile, the SA (Shuffle Attention) mechanism is introduced at the end of the backbone network to enhance the extraction of valid information and improve detection accuracy. Optimizing the bounding box loss function to WIoU v3 (Wise-IoU v3) accelerates the convergence speed of the network. The proposed network model was tested using the ground-planted strawberry dataset, and the experimental results show that the P, R, and mAP@.5 of the improved network model are 82.8 %, 82.1 %, and 86.7 %, respectively, which are improved by 5.0 %, 1.7 %, and 2.2 %, respectively, compared with the YOLOv7 model. By comparing with other classical target detection algorithms, it is verified that the proposed model has advantages in terms of accuracy and model size. Finally, the DSW-YOLO model was deployed on Jetson Xavier NX of a strawberry picking robot, and visual recognition tests were conducted in field. The detection time of a single image is 92 ms, which satisfies the speed requirement of robot picking, and the detection results verify that the DSW-YOLO network model can quickly and accurately detect ripe strawberry fruits under different occlusion levels in a complex field environment, which provides a theoretical basis for the picking robot to achieve accurate positioning and efficient picking.

Adversarial Robust Aerial Image Recognition Based on Reactive-Proactive Defense Framework with Deep Ensembles

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.