Search Results (177)

Background/Objectives: Artificial intelligence (AI) is beginning to be applied in wound ballistics, showing preliminary potential to improve the accuracy and objectivity of forensic analyses. This review explores the current state of AI applications in forensic firearm wound analysis, emphasizing its potential to address challenges such as subjective interpretations and data heterogeneity. Methods: A systematic review adhering to PRISMA guidelines was conducted using databases such as Scopus and Web of Science. Keywords focused on AI and GSW classification identified 502 studies, narrowed down to 4 relevant articles after rigorous screening based on inclusion and exclusion criteria. Results: These studies examined the role of deep learning (DL) models in classifying GSWs by type, shooting distance, and entry or exit characteristics. The key findings demonstrated that DL models like TinyResNet, ResNet152, and ConvNext Tiny achieved accuracy ranging from 87.99% to 98%. Models were effective in tasks such as classifying GSWs and estimating shooting distances. However, most studies were exploratory in nature, with small sample sizes and, in some cases, reliance on animal models, which limits generalizability to real-world forensic scenarios. Conclusions: Comparisons with other forensic AI applications revealed that large, diverse datasets significantly enhance model performance. Transparent and interpretable AI systems utilizing techniques are essential for judicial acceptance and ethical compliance. Despite the encouraging results, the field remains in an early stage of development. Limitations highlight the need for standardized protocols, cross-institutional collaboration, and the integration of multimodal data for robust forensic AI systems. Future research should focus on overcoming current data and validation constraints, ensuring the ethical use of human forensic data, and developing AI tools that are scientifically sound and legally defensible. Full article

Pedestrian detection systems are widely used in safety-critical domains such as autonomous driving, where deep neural networks accurately perceive individuals and distinguish them from other objects. However, their vulnerability to backdoor attacks remains understudied. Existing backdoor attacks, relying on unnatural digital perturbations or explicit patches, are difficult to deploy stealthily in the physical world. In this paper, we propose a novel backdoor attack method that leverages real-world occlusions (e.g., backpacks) as natural triggers for the first time. We design a dynamically optimized heuristic-based strategy to adaptively adjust the trigger’s position and size for diverse occlusion scenarios, and develop three model-independent trigger embedding mechanisms for attack implementation. We conduct extensive experiments on two different pedestrian detection models using publicly available datasets. The results demonstrate that while maintaining baseline performance, the backdoored models achieve average attack success rates of 75.1% on KITTI and 97.1% on CityPersons datasets, respectively. Physical tests verify that pedestrians wearing backpack triggers could successfully evade detection under varying shooting distances of iPhone cameras, though the attack failed when pedestrians rotated by 90°, confirming the practical feasibility of our method. Through ablation studies, we further investigate the impact of key parameters such as trigger patterns and poisoning rates on attack effectiveness. Finally, we evaluate the defense resistance capability of our proposed method. This study reveals that common occlusion phenomena can serve as backdoor carriers, providing critical insights for designing physically robust pedestrian detection systems. Full article

Objective: This study aims to investigate the effect of shot selection on the muscle coordination characteristics during basketball shooting. Methods: A three-dimensional motion capture system, force platform, and wireless surface electromyography (sEMG) were used to simultaneously collect shooting data from 14 elite basketball players. An inverse mapping model of sEMG signals and spinal α-motor neuron pool activity was developed based on the Debra muscle segment distribution theory. Non-negative matrix factorization (NMF) and K-means clustering were used to extract muscle coordination features. Results: (1) Significant differences in spinal segment activation timing and amplitude were observed between stationary and jump shots at different distances. In close-range stationary shots, the C5-S3 segments showed higher activation during the TP phase and lower activation during the RP phase. For mid-range shots, the C6-S3 segments exhibited greater activation during the TP phase. In long-range shots, the C7-S3 segments showed higher activation during the TP phase, whereas the L3-S3 segments showed lower activation during the RP phase (p < 0.01). (2) The spatiotemporal structure of muscle coordination modules differed significantly between stationary and jump shots. In terms of spatiotemporal structure, the second and third coordination groups showed stronger activation during the RP phase (p < 0.01). Significant differences in muscle activation levels were also observed between the coordination modules within each group in the spatial structure. Conclusion: Shot selection plays a significant role in shaping neuromuscular control strategies during basketball shooting. Targeted training should focus on addressing the athlete’s specific shooting weaknesses. For stationary shots, the emphasis should be on enhancing lower limb stability, while for jump shots, attention should be directed toward improving core stability and upper limb coordination. Full article

The increasing frequency of mass shootings at public events and public buildings underscores the limitations of traditional surveillance systems, which rely on human operators monitoring multiple screens. Delayed response times often hinder security teams from intervening before an attack unfolds. Since firearms are rarely seen in public spaces and constitute anomalous observations, firearm detection can be considered as an anomaly detection (AD) problem, for which one-class classifiers (OCCs) are well-suited. To address this challenge, we propose a holistic firearm detection approach that integrates OCCs with visual hand-held gun features and human pose estimation (HPE). In the first stage, a variational autoencoder (VAE) learns latent representations of firearm-related instances, ensuring that the latent space is dedicated exclusively to the target class. Hand patches of variable sizes are extracted from each frame using body landmarks, dynamically adjusting based on the subject’s distance from the camera. In the second stage, a unified feature vector is generated by integrating VAE-extracted latent features with landmark-based arm positioning features. Finally, an isolation forest (IFC)-based OCC model evaluates this unified feature representation to estimate the probability that a test sample belongs to the firearm-related distribution. By utilizing skeletal representations of human actions, our approach overcomes the limitations of appearance-based gun features extracted by camera, which are often affected by background variations. Experimental results on diverse firearm datasets validate the effectiveness of our anomaly detection approach, achieving an F1-score of 86.6%, accuracy of 85.2%, precision of 95.3%, recall of 74.0%, and average precision (AP) of 83.5%. These results demonstrate the superiority of our method over traditional approaches that rely solely on visual features. Full article

In nature, orchid seed germination is extremely low, making in vitro asymbiotic seed germination essential for the propagation and conservation of endangered Vanda coerulea. This study optimized a micropropagation protocol and evaluated the genetic homogeneity of regenerated orchids. The synergistic effect of kinetin (KN) with auxins in the Mitra (M) medium best supported protocorm formation and seedling development. The highest shoot multiplication (5.62 ± 0.09) was achieved with 1.2 mg L⁻¹ KN and 0.6 mg L⁻¹ IBA (indole-3-butyric acid) in the medium. Enhanced leaf production (4.81 ± 0.37) was observed when 3.2 mg L⁻¹ KN was combined with 1.8 mg L⁻¹ IAA (indole-3-acetic acid), while root development was superior when 3.2 mg L⁻¹ KN together with 2.4 mg L⁻¹ IAA was incorporated in the medium. Anatomical sections confirmed well-developed leaf and root structures. Genetic fidelity assessment using random amplified polymorphic DNA (RAPD), inter-simple sequence repeat (ISSR), inter-primer binding site (iPBS), and start codon targeted (SCoT) markers revealed 97.17% monomorphism (240/247 bands) and low Nei’s genetic distances (0.000–0.039), indicating high similarity among the regenerants. Dendrogram clustering was supported by a high cophenetic correlation coefficient (CCC = 0.806) and strong resolution in Principal Coordinate Analysis (PCoA) (44.03% and 67.36% variation on the first two axes). The Mantel test revealed a significant correlation between both ISSR and SCoT markers with the pooled marker data. Flow cytometry confirmed the genome stability among the in vitro-propagated orchids, with consistently low CV (FL2-A) values (4.37–4.94%). This study demonstrated the establishment of a reliable in vitro protocol for rapidly propagating genetically identical V. coerulea via asymbiotic seed germination. Full article

The peptide ligands of the CLAVATA3/EMBRYO SURROUNDING REGION-RELATED (CLE) family have been previously identified as essential signals for both short- and long-distance communication in plants, particularly during stem cell homeostasis, cell fate determination, and growth and development. To date, most studies on the CLE family have focused on model plants and especially those involving stem and apical meristems. Relatively little is known about the role of CLE peptides in tall trees and other plant meristems. In this review, we summarize the role of CLE genes in regulating plant Root Apical Meristem (RAM), Shoot Apical Meristem (SAM), Procambium, Leaf and Floral Meristem (FM), as well as their involvement in multiple signaling pathways. We also highlight the evolutionary conservation of the CLE gene family and provide a comprehensive summary of its distribution across various plant developmental tissues. This paper aims to provide insights into novel regulatory networks of CLE in plant meristems, offering guidance for understanding intercellular signaling pathways in forest trees and the development of new plant organs. Full article

To address the low estimation accuracy of deep learning-based crop yield image recognition methods under untrained shooting distances, this study proposes a shooting distance adaptive crop yield estimation method by fusing RGB and depth image information through multi-modal data fusion. Taking strawberry fruit fresh weight as an example, RGB and depth image data of 348 strawberries were collected at nine heights ranging from 70 to 115 cm. First, based on RGB images and shooting height information, a single-modal crop yield estimation model was developed by training a convolutional neural network (CNN) after cropping strawberry fruit images using the relative area conversion method. Second, the height information was expanded into a data matrix matching the RGB image dimensions, and multi-modal fusion models were investigated through input-layer and output-layer fusion strategies. Finally, two additional approaches were explored: direct fusion of RGB and depth images, and extraction of average shooting height from depth images for estimation. The models were tested at two untrained heights (80 cm and 100 cm). Results showed that when using only RGB images and height information, the relative area conversion method achieved the highest accuracy, with R² values of 0.9212 and 0.9304, normalized root mean square error (NRMSE) of 0.0866 and 0.0814, and mean absolute percentage error (MAPE) of 0.0696 and 0.0660 at the two untrained heights. By further incorporating depth data, the highest accuracy was achieved through input-layer fusion of RGB images with extracted average height from depth images, improving R² to 0.9475 and 0.9384, reducing NRMSE to 0.0707 and 0.0766, and lowering MAPE to 0.0591 and 0.0610. Validation using a developed shooting distance adaptive crop yield estimation platform at two random heights yielded MAPE values of 0.0813 and 0.0593. This model enables adaptive crop yield estimation across varying shooting distances, significantly enhancing accuracy under untrained conditions. Full article

Tower bolts play a crucial role as connecting components in wind turbines and are of great interest for health monitoring systems. Non-contact monitoring techniques offer superior efficiency, convenience, and intelligence compared to contact-based methods. However, the precision and robustness of the non-contact monitoring process are significantly impacted by suboptimal lighting conditions within the wind turbine tower. To address this problem, this article proposes an automated detection method for the bolt detachment of wind turbines in low-light scenarios. The approach leverages the deep convolutional generative adversarial network (DCGAN) to expand and augment the small-sample bolt dataset. Transfer learning is then applied to train the Zero-DCE++ low-light enhancement model and the bolt defect detection model, with the experimental verification of the proposed method’s effectiveness. The results reveal that the deep convolutional generative adversarial network can generate realistic bolt images, thereby improving the quantity and quality of the dataset. Additionally, the Zero-DCE++ light enhancement model significantly increases the mean brightness of low-light images, resulting in a decrease in the error rate of defect detection from 31.08% to 2.36%. In addition, the model’s detection performance is affected by shooting angles and distances. Maintaining a shooting distance within 1.6 m and a shooting angle within 20° improves the reliability of the detection results. Full article

Background: Handedness dominance can be observed in the tactical aspects of a handball match geometry. Therefore, this study aimed to examine the asymmetry between shooting velocity and distance in left- and right-handed handball players, and also to see if there is a difference between scored and missed shots. Methods: The data were obtained from players participating in the EHF European Championship 2024, held in Germany. Results: In this study, 238 players were analyzed during the whole championship. They were divided into two groups: left- (N = 112) and right-handed players (N = 126). A total of 5710 shots taken by the players were collected and analyzed. The results show that the left-handed players had a higher score percentage (63.08%) than the right-handed players (57.86%). The right-handed players shot at a higher velocity (101.38 ± 18.00 km/h) than the left-handed players (99.36 ± 18.89 km/h) (p < 0.001). A similar difference was observed in the distance of the shots (7.61 ± 2.23 m; and 7.42 ± 2.59 m, respectively) (p < 0.001). The distance of the shots differed between the scored and missed shots (right-handed, p < 0.001; left-handed, p < 0.04). Conclusions: These findings suggest that an asymmetry in left- and right-handed players is present for both parameters. Also, the higher efficiency of the right side of a handball team could lead to asymmetry in the geometry of a handball match. Full article

Unmanned aerial vehicle (UAV) imagery often suffers from significant object scale variations, high target density, and varying distances due to shooting conditions and environmental factors, leading to reduced robustness and low detection accuracy in conventional models. To address these issues, this study adopts DGBL-YOLOv8s, an improved object detection model tailored for UAV perspectives based on YOLOv8s. First, a Dilated Wide Residual (DWR) module is introduced to replace the C2f module in the backbone network of YOLOv8, enhancing the model’s capability to capture fine-grained features and contextual information. Second, the neck structure is redesigned by incorporating a Global-to-Local Spatial Aggregation (GLSA) module combined with a Bidirectional Feature Pyramid Network (BiFPN), which strengthens feature fusion. Third, a lightweight shared convolution detection head is proposed, incorporating shared convolution and batch normalization techniques. Additionally, to further improve small object detection, a dedicated small-object detection head is introduced. Results from experiments on the VisDrone dataset reveal that DGBL-YOLOv8s enhances detection accuracy by 8.5% relative to the baseline model, alongside a 34.8% reduction in parameter count. The overall performance exceeds most of the current detection models, which confirms the advantages of the proposed improvement. Full article

The long-term spatial–temporal variation in shoot blight of larch in China has not yet been clearly defined, and the mechanisms behind its long-distance spread remain unknown. This study, based on the historical occurrence dataset of shoot blight of larch in China, used spatial statistical analysis to describe the spatial changes in the disease across five stages since 1973. Subsequently, the study utilized Geo Detector and Random Forest models to investigate the relationship between the spread and occurrence of shoot blight of larch and seven influencing factors. The results revealed the following: (1) The spread of shoot blight of larch in China exhibits significant directionality, with the affected regions distributed along a northeast–southwest axis, and the epicenter of the spread is shifting southwestward; (2) Shandong and Jilin provinces served as the initial introduction points for shoot blight of larch, with most infected counties in other provinces experiencing outbreaks between 1989 and 1996, accompanied by a noticeable spread to neighboring provinces; (3) the occurrence of shoot blight of larch demonstrates a significant positive spatial clustering effect, forming a monocentric “core–periphery” structure centered in Liaoning Province, where kernel density values decrease gradually outward from the core. Geo Detector identified “seedling planting area” as a potential spatial driving factor for the disease. These findings underscore the critical influence of the combined effects of human activities and natural factors in shaping the spatiotemporal distribution patterns of shoot blight of larch. Full article

In recent years, the continuous advancement of deep learning has led to significant progress in object-detection technology for remote-sensing images. However, most current detection methods passively perform detection on the input image without considering the relationship between imaging configurations and detection-algorithm performance. Therefore, when factors such as poor lighting conditions, extreme shooting angles, or long acquisition distances degrade image quality, the passive detection framework limits the effectiveness of the current detection algorithm, preventing it from completing the detection task. To address the limitations above, this paper proposes an active object-detection (AOD) method based on deep reinforcement learning, taking adaptive brightness and collection position adjustments as examples. Specifically, we first established an end-to-end network structure to generate attribute control policies. Then, we designed a reward function suitable for remote-sensing images based on the degree of improvement in detection performance. Finally, we propose a new viewpoint-management method in this paper, which is successfully implemented by a training method of long-term Prioritized Experience Replay (LPER), which significantly reduces the accumulation of negative and repetitive samples and improves the success rate of the AOD algorithm for remote-sensing images. The experiments on two public datasets have fully demonstrated the effectiveness and advantages of the algorithm proposed in this paper. Full article

Team dynamics significantly influence the outcomes of modern football matches. This study employs an information-theoretical approach, specifically causal emergence, combined with graph theory to explore how team-level dynamics arise from complex interactions among players, utilizing tracking data from 34 J-League matches. We focused on how collective behaviors arise from the interdependence of individual actions, examining team coordination and dynamics through player positions and movements to identify emergent properties. Specifically, we selected relative distance to the field’s center, center of mass (CoM) and clustering coefficients based on velocity similarity and inverse distance as macroscopic features to capture the key aspects of team structure, coordination, and spatial relationships. Relative distance and CoM represent the collective positioning of the team, while clustering coefficients provide insights into localized cooperation and movement similarity among the players. The results indicate that average causal emergence with relative distance and CoM as a macroscopic feature across entire games shows a strong correlation with differences in ball possession rate between home and away teams. In contrast, clustering coefficients based on inverse distance and velocity similarity showed moderate to weak correlations with ball possession rate, indicating that these metrics may capture localized interactions that are less directly tied to team-level emergent behavior compared to CoM. Additionally, relative distance and CoM as macroscopic features yield higher causal emergence in attacking phases than in defending phases before shooting, suggesting that the collective positioning of players may play a more significant role in facilitating successful attacks than in defensive stability. This study offers a novel perspective on team coordination in football, suggesting that effective team coordination may be characterized by emergent patterns arising from collective positioning. These findings have practical implications for understanding coordinated team behaviors and inform coaching and performance analysis focused on enhancing team dynamics. Full article

Bamboo, characterized by its high growth speed and short maturation period, occupies 0.875% of the global forest area and significantly contributes to terrestrial carbon cycling. The state of shoot growth can essentially indicate a bamboo forests’ health and productivity. This study explored the potential of backpack laser scanning (BLS) for monitoring the growth of Moso bamboo shoots (Phyllostachys edulis), a key economic species in subtropical China. Initially, the accuracy of BLS in extracting attributes of bamboo and shoots (including diameter at breast height (DBH), height, and real-world coordinates) was validated. An optimized method was developed to address the lower precision of BLS in extracting the DBH for thinner species. Subsequently, this research analyzed the impact of spatial structure and other indicators on shoot emergence stage and growth rate using a random forest model. The results indicate that BLS can accurately extract Moso bamboo and shoot height (RMSE = 0.748 m) even in dense bamboo forests. After optimization, the error in DBH extraction significantly decreased (RMSE = 0.835 cm), with the average planar and elevation errors for Moso bamboo being 0.227 m and 0.132 m, respectively. The main indicators affecting the coordinate error of Moso bamboo were the distance to the start (DS) and the distance to the trajectory (DT). The emergence time of shoots was mainly influenced by the surrounding Moso bamboo quantity, with the leaf area index (LAI) and competition index (CI) positively related to the growth rate of shoots. The importance ranking of spatial structure for the carbon storage of shoots was similar to that of the growth rate of shoots, with both identifying LAI as the most significant indicator. This study has validated the value of BLS in monitoring the growth of shoots, providing a theoretical support for the sustainable management and conservation of bamboo forests. Full article

An arrow’s attack angle continuously changes during its flight, which affects the position of the aerodynamic pressure center. To account for the offset between the aerodynamic pressure center and the center of mass of an arrow in recurve bow shooting, two equations for describing the variation of the aerodynamic pressure center with the attack angles were fitted via CFD simulation. On this basis, a new theoretical aerodynamic model was developed by integrating the above equations with the current model to predict the flight along the outer ballistic trajectory more accurately than ever. With regard to actual archery competition occasions, the distance, initial velocity, and attack angle were set to 70 m, 57 m/s, and −3° to 3°, respectively; the attitude and trajectory of the arrow flying details under background wind, such as crosswind, headwind, and tailwind, were numerically analyzed to reveal the accuracy deviation mechanism. A comparison was conducted with previous models, indicating that the model proposed in this study achieved improvements in accuracy of 15% under crosswind conditions and 8% under headwind/tailwind conditions. The results could, from a fluid physics perspective, provide valuable recommendations not only for archers and coaches but also for arrow manufacturers. Full article