Search Results (8,743)

The increasing demand for intelligent transportation systems (ITS) and advanced driver assistance system (ADAS) significantly demands a real-time and robust perception to recognize road-side obstacles in varying different weather settings. This paper presents DriveEdgeAI, a lightweight YOLO11 based embedded deep learning framework for efficient road anomaly detection with the emphasis on potholes, speed bumps and relevant traffic sign detection. We have prepared a custom dataset consisting of 17,061 annotated images to train and test the model under different lighting conditions, weather conditions, and roads configurations. The proposed system also managed to demonstrate good convergence and generalization with a precision@50 of 95.8%, recall@50 of 89.7%, mAP@50 of 95.4%, surpassing previous YOLO versions. The stability and robustness of the model at different thresholds were also substantiated by Precision-Recall and F1-Confidence analyses. DriveEdgeAI was also deployed on a number of edge devices, such as Jetson Nano, Raspberry Pi 5, Intel Movidius VPU and Hailo-8L NPU respectively reaching 9.5 FPS/W and 28.5 FPS for the Raspberry Pi 5 + Hailo-8L version. From these results, one can conclude that DriveEdgeAI is an energy-efficient and scalable solution for real-world ADAS applications. Full article

Conventional defect inspection for warp-knitted lace relies on manual work and negative-sample-based training, resulting in low efficiency, frequent false detections and poor adaptability. This study presents a novel AI visual inspection system centered on positive-sample learning, which is built upon a five-layer 5G + Industrial Internet distributed architecture. Supported by modified looms, high-precision imaging devices and an optimized YOLOv5s model, the system accomplishes intelligent defect detection. A positive-sample self-learning paradigm and dual-model collaboration mechanism are proposed to reduce the demand for negative samples and cut labeling expenses. The integration of CBAM, FPN + PAN structure, self-supervised learning and hybrid loss further strengthens the recognition performance for subtle defects under complex patterns. Industrial tests show that the system reaches a grid-level classification accuracy of 95% and a frame-level detection rate over 98%, with a detection speed of 30 m/min. It reduces labor costs and product reject rates by 40% and 30% correspondingly while running stably in real production. This method breaks the constraints of traditional training modes, provides a scalable intelligent solution for the digital upgrading of the warp-knitted lace industry, and promotes the high-quality development of textile manufacturing. Full article

This systematic review examined the validity and reliability of wearable inertial sensor systems to quantify spatiotemporal gait parameters in post-stroke adults, a population in which gait asymmetry and altered motor control challenge accurate measurement. Sixteen studies involving 300 participants were included. Spatial parameters gait speed, cadence, and step/stride length showed consistently good-to-excellent agreement with reference systems (ICC 0.85–0.98; 95% LoA ±0.03–0.08 m/s for gait speed, ±4–10 steps/min for cadence, and ±3–8 cm for step/stride length) and high test–retest reliability. Temporal parameters demonstrated greater heterogeneity, with larger errors and lower concordance (ICC 0.40–0.85; LoA ±0.04–0.12 s), particularly for swing time (ICC 0.40–0.70; LoA up to ±0.15 s). Paretic-side measurements showed 10–20% lower concordance and 30–50% wider limits of agreement compared with the non-paretic side, although within-subject reliability remained moderate to high. No consistent influence of sensor number on measurement accuracy was observed. Overall, wearable inertial sensors provide robust estimates of spatial gait parameters, whereas temporal outcomes especially swing time remain limited due to challenges in gait event detection under stroke-related biomechanical alterations. These findings highlight the need for standardized protocols and improved algorithms to enhance comparability across studies and support broader clinical adoption. Full article

Real-time fruit maturity detection in unstructured orchards remains challenging because of variable illumination, fruit occlusion, complex backgrounds, and the limited computing capacity of edge devices. To address these challenges, this study proposes WSS-YOLO, a lightweight detection framework based on YOLOv11n for quince maturity detection. The model introduces WaveletPool to reduce texture loss during downsampling, adopts a GSConv-based Slim-neck to improve feature fusion with lower computational cost, and integrates SimAM to enhance discriminative fruit-region responses without adding trainable parameters. Experiments on a multi-scenario quince maturity dataset show that WSS-YOLO achieves 86.4% precision, 87.5% recall, and 93.4% mAP@0.5, improving the YOLOv11n baseline by 2.3, 1.7, and 2.5 percentage points, respectively. The model contains only 2.23 M parameters and requires 4.1 G FLOPs. Deployment on the NVIDIA Jetson Orin Nano achieved a real-time speed of 23.0 FPS, suggesting a favorable trade-off between detection accuracy and computational efficiency under the tested conditions. Full article

Poultry meat producers face pressure to ensure both safety and efficiency as food systems expand in scale and complexity. However, existing diagnostic tools often force a trade-off between speed, accuracy, and cost, leaving limited options for real-time, on-site pathogen detection. This study evaluated the performance of a novel point-of-use rapid diagnostic tool combining filtration and lyophilized isothermal amplification for detecting and semi-quantifying Salmonella in chicken carcass rinse samples. Targeted DNA (invA) was amplified at 65 °C for 60 min, and fluorescence was measured over time. Salmonella was successfully detected in all spiked samples (1.0- to 7.5-log10 CFU/mL; n = 57). An association was observed between the isothermal detection speed data and Salmonella load in the sample (r² = 0.90; p < 0.001), particularly at concentrations ≥ 1.5-log10 CFU/mL. Repeatable data were obtained across three operators and samples of multiple origins (Georgia, Illinois, Nebraska; 0.16-log10; 95%CI: 0.11–0.21). Importantly, complete inclusivity for all tested Salmonella serotypes (n = 46/46) and exclusivity against non-target organisms (n = 0/37) was validated in this study. In conclusion, the technological combination of filtration and lyophilized isothermal amplification enabled the point-of-use detection and quantification of Salmonella in chicken carcass samples within 60 min and minimal lab infrastructure. Full article

Image-based measurement of chopped maize stalk length remains difficult because the fragments are often slender, curved, touching, or partly overlapped. Bounding-box dimensions are therefore not reliable for length estimation, and manual measurement is too slow for repeated quality assessment. In this study, we developed a YOLO11n-based visual framework for measuring chopped maize stalk length under fixed imaging conditions. The dataset contained 1127 images collected on a laboratory platform and covered stalk lengths of 10–150 mm, different moisture states, and isolated, touching, and overlapping arrangements. To obtain more stable regions of interest, the YOLO11n detector was modified with large separable kernel attention (LSKA), a lightweight cross-scale decoupled detection (LSCD) head, and Wise intersection over union version 3 (WIoU v3). The detected stalk regions were then processed by local segmentation, morphological refinement, skeleton extraction, longest-path calculation, and washer-based scale conversion. The modified detector reached 94.8% precision, 90.4% recall, 96.5% mAP@0.5, and 71.1% mAP@0.5:0.95, with a detector inference speed of 174 FPS. In the length-measurement test, the mean relative errors were 5.8%, 8.3%, and 10.4% for the <40 mm, 40–80 mm, and >80 mm groups, respectively. Across all evaluated fragments, the complete pipeline produced an MAE of 6.0 mm, an RMSE of 9.4 mm, and a mean relative error of 8.2%. The framework therefore provides a practical way to measure chopped maize stalk length under controlled imaging conditions, although long, curved, and cluttered fragments still caused most of the remaining errors. Full article

This work addresses the design and implementation of an automated system for the handling and transportation of parts, integrating speed sensors, an optimized PID controller, an HMI interface, and an industrial robotic system. The speed sensors, powered by 5 V DC, enable continuous measurement of the conveyor belt’s speed and direction of rotation, providing the feedback signal required for the control loop. The core element of the system is the implementation of a PID controller applied to a direct current motor responsible for driving the conveyor belt. This controller regulates the motor speed by analyzing the error between the reference speed and the measured speed, using proportional, integral, and derivative actions to improve system stability, reduce steady-state error, and minimize oscillations. The application of PID control makes it possible to achieve an appropriate dynamic response, ensuring accuracy and reliability in the transportation process. System monitoring and operation are carried out through a human–machine interface (HMI) developed in LOGO Web Editor, which communicates with the PLC (LOGO V8) to visualize and control the status of the conveyor belt, sensors, and control elements in real time. This interface facilitates interaction between the operator and the system, allowing both virtual and physical operation. In addition, RAPID programming is used to control the IRB 14000 industrial robot, enabling the reading of PLC signals and the execution of coordinated trajectories between both arms. The operating sequence includes picking up a part with the left arm, placing it on the conveyor belt, and, after detection by sensors and PLC control, subsequent manipulation by the right arm to a specific point. Finally, both arms return to their original position, ensuring synchronized and collision-free operation. Lastly, this work integrates scientific knowledge related to the modeling, analysis, and control of dynamic systems, particularly in the implementation of closed-loop PID control optimized using genetic algorithms. This control is applied directly to an embedded system through the use of an Arduino board as the processing and control platform. Likewise, technological knowledge associated with industrial automation, PLC programming, HMI development, and industrial robotics is incorporated. The convergence of these scientific and technological approaches results in a comprehensive and compelling project that demonstrates the practical application of theoretical concepts in a functional automated system representative of real industrial environments. Full article

Magnetostrictive patch transducers (MPTs) are highly efficient for generating and detecting ultrasonic waves for non-destructive evaluation (NDE), though their use on cementitious media and fibre-reinforced concrete has not yet been investigated. In this study, a COMSOL simulation, validated with laser-Doppler vibrometry, was first used to quantify patch deformation for use in subsequent simulation of wave propagation in samples. The MPT system was then validated on thin glass plates, producing tunable A₀, S₀, and SH₀ modes through frequency-wavelength matching. In cementitious mortar plates, SH₀ and SH₁ modes were demonstrated experimentally for the first time using MPTs. The validated COMSOL model was then used to interpret complex signals in quasi-plate and half-space cementitious mortar prisms, showing that MPTs generate Rayleigh, bulk SH, and surface-skimming SH modes. In steel fibre-reinforced concrete, surface-skimming SH wave speed correlated with increases in breaking strength even in the presence of surface features such as notches. Notably, Rayleigh wave speeds could not be measured in the presence of surface features, and the Rayleigh velocities measured in the same sample, but not in the local tested area did not correlate with SH speed. This behaviour is likely due to the non-uniform distribution of material constituents, including fibre-reinforcement and coarse aggregate, combined with the different propagation paths and depth sensitivities of the reported wave modes. Overall, racetrack-coil MPTs enable multimodal inspection of cementitious media, providing information on the presence of geometric features and material properties. Full article

Accurate and rapid detection of Volvariella volvacea (straw mushroom) fruiting bodies at harvestable maturity is a critical prerequisite for automated industrial cultivation. However, existing detection methods often yield high false-negative and false-positive rates when processing a small-scale, densely distributed, and heavily occluded targets against complex straw substrate backgrounds. Furthermore, these methods frequently struggle to balance the competing requirements of architectural efficiency (such as parameter volume and computational complexity) and real-time performance for edge computing. To address these challenges, this study proposes a YOLO11n-CPDM, a lightweight detection model based on an improved YOLO11n architecture. The model incorporates synergistic optimizations across feature extraction, fusion, and reconstruction. First, a Dual Coordinate Attention Feature Extraction mechanism is integrated into the C3k2 bottleneck blocks of the backbone network. This enhances target perception in complex, occluded environments by concurrently modeling global context and local salient features. Second, within the neck network, the standard attention module is replaced with the PnPNystraAttention module, coupled with the DySample dynamic upsampling operator. This modification strengthens contextual relationships among multi-scale features and improves spatial consistency during reconstruction while preserving linear computational complexity. Finally, the detection head is optimized using MBConv blocks based on an inverted residual structure to minimize parameter volume. Experimental results on a custom V. volvacea dataset demonstrate that the proposed YOLO11n-CPDM model achieves significant performance gains, with Precision (P), Recall (R), and Mean Average Precision (mAP50) reaching 86.8%, 87.5%, and 88.4%, respectively. These figures represent improvements of 2.7, 3.0, and 3.2 percentage points over the baseline YOLO11n model. Additionally, the model size is reduced to 4.8 MB (a 12.7% decrease), while achieving inference speeds of 42.7 FPS on Jetson AGX Orin and 21.2 FPS on Jetson Nano, outperforming the baseline model on both embedded platforms. Consequently, the proposed model effectively enhances detection performance in complex environments while maintaining excellent lightweight characteristics and deployment flexibility, providing a solid technical foundation for intelligent perception and automated harvesting of V. volvacea. Full article

Burdur Lake is a saline lake in southwestern Türkiye that has undergone significant hydrological changes in recent decades. Water levels dropped by about 18 m, and the surface area became 75 km² smaller between 1980 and 2022. Trend analyses showed a strong downward trend in lake levels (p < 0.01), while air temperatures increased significantly. Wind speed and relative humidity decreased, and no significant trend in precipitation was detected. Analysis of the normalized difference vegetation index (NDVI) values showed that the vegetation density increased during the same period. Correlation results indicated no link with precipitation and water levels, but a strong negative relationship of water levels with temperature and NDVI was found. Overall, the decline in lake level cannot be explained solely by climatic changes; anthropogenic effects are also highly influential. Full article

Radio-frequency identification (RFID)-based asset tracking in cabinet environments often encounters unpredictable detection caused by multipath fading, metal-induced interference, and tag placement sensitivity, which can render single-band systems unreliable under real-world conditions. This paper proposes a dual-band detection approach combining 915 MHz and 2.45 GHz to address these challenges through frequency diversity. Unlike designs confined to closely spaced UHF bands, this method uses a larger spectral gap to benefit from uncorrelated fading and distinct propagation properties. Theoretical analysis shows that dual-band detection significantly reduces joint failure probability under independent fading. The proposed framework is implemented using commercially available passive UHF tags at 915 MHz and an active RFID tag/reader at 2.45 GHz. The two systems are operated sequentially along the same guided scan path, and their detected tag-ID sets are combined offline using an OR-fusion rule without hardware-level synchronization. Across trials with varied scan speeds, power levels, reader distances, and tag placements, single-band detection fell below 50% under double-speed scanning at 200 cm, while the dual-band method remained above 70% and, in many cases, reached 100% reliability. Performance trends are further analyzed across individual scenarios, showing that 2.45 GHz links are less affected by metallic shadowing at close range, whereas 915 MHz links maintain more stable detection at longer distances. These findings are discussed in terms of deployment feasibility, indicating that the additional hardware and configuration requirements are offset by the measurable improvement in detection consistency, making the approach applicable for inventory tracking in logistics, warehousing, and industrial automation. Full article

To evaluate the effects of swimming training on growth, swimming performance, and osmoregulatory-related indices in juvenile coho salmon, freshwater-reared fish were subjected to current of 1 body length per second (BL·s⁻¹) from December 2024 to April 2025. Fork length, body weight, condition factor, insulin-like growth factor-1 (IGF-1), and gill and intestinal Na⁺/K⁺-ATPase (NKA) protein abundance were measured monthly, and critical swimming speed (U_crit) was evaluated after one month of training. Trained fish showed greater fork length in March and higher body weight in March and April than controls. The condition factor was higher in trained fish in February and March, but declined during spring smolt development. Swimming capacity was enhanced by training, as indicated by significantly higher U_crit. Mean IGF-1 levels did not differ between groups, but IGF-1 correlated positively with body size only in trained fish. No significant training effect was detected for either gill or intestinal NKA protein abundance, although gill NKA increased significantly in April, likely reflecting seasonal smoltification. In addition, IGF-1 was significantly correlated with gill NKA in trained fish in March. Collectively, these results indicate that sustained swimming training improves growth and swimming performance and may enhance associations among measured physiological variables during smoltification in juvenile coho salmon. Full article

Accurate detection of field insect pests remains a significant challenge for precision agriculture due to the elongated and variable morphology of the target organisms, their frequent resemblance to complex background textures, and the long-tail distribution of species in natural datasets. While deep convolutional neural networks (CNNs) have advanced the field, they are often constrained by a limited effective receptive field and the entanglement of semantic and spatial features, which can lead to elevated false-positive rates and missed detections for low-contrast or rare targets. This paper introduces a novel detection framework that integrates state space modeling with multi-stream feature disentanglement to address these limitations. First, a visual state space module is employed as the backbone feature extractor, enabling the establishment of a global receptive field with linear computational complexity and thereby improving the perception of long-range morphological structures. Second, a Topological Feature Disentanglement Pyramid Network is proposed. This architecture explicitly separates feature representations into semantic and spatial streams and recombines them through graph convolutional interactions, which serves to suppress background interference and enhance localization precision. A meta-auxiliary detection head, active only during training, is introduced to amplify supervision signals for hard, low-contrast samples via adversarial gradient modulation. Furthermore, an implicit neural radiance field augmentation pipeline is used to generate physically consistent synthetic views of underrepresented pest classes, mitigating the negative effects of long-tail data distributions. Experimental evaluations on the public BAU-Insectv2 benchmark demonstrate that the proposed method achieves a mean average precision (mAP@0.5) of 81.8%, representing a 4.4-percentage-point improvement over a comparable baseline, while maintaining a compact parameter count of 2.33 M and an inference speed of 178.6 FPS. The framework exhibits particular efficacy in detecting elongated, minute, and rare pests, suggesting a promising technical approach for real-time, field-based pest surveillance in precision agriculture. Full article

In this work, a lightweight framework enabled by the modified YOLOv11s-seg model for tea shoot detection and plucking point localization is proposed. Detecting tea shoots and localizing plucking points with higher accuracy generally require larger model size and more model parameters, making it difficult to balance accuracy and lightweighting. To overcome this limitation, a modified lightweight YOLOv11s-seg model is developed. First, the multi-scale edge information enhancement is introduced into the conventional YOLOv11s-seg to extract edge feature better and improve the detection accuracy of tea shoots. Meanwhile, context anchor attention is utilized to modify the cross stage partial spatial attention module in a backbone network to improve the detection capability for small objects. Moreover, the detail calibration reconstruction feature pyramid network is proposed. It utilizes spatial and contextual semantic information to reconstruct and calibrate features in key regions, enhancing the capability for object fusion and recognition at various scales. Furthermore, with the modified model performing instance segmentation to acquire the contour of each tea shoot, the coordinates of the three lowest pixel points in the contour are captured to localize the plucking point based on the average coordinates. In addition, the layer-adaptive magnitude-based pruning (LAMP) method is used to lighten the model. The experimental results show that the LAMP-pruned modified YOLOv11s-seg model with a speedup ratio of 1.5 achieves a mAP@0.5 of 86.5% for tea shoot detection, exhibiting a 4.7 percentage point improvement over the conventional YOLOv11s-seg model. Moreover, it exhibits an accuracy of 81.9% for plucking point localization on the validation and test subsets with 232 images in total, and its number of parameters, model size and floating point operations (FLOPs) separately achieve reductions of 67.3%, 66.2%, and 24.9% over the conventional model as well. Therefore, the proposed LAMP-pruned modified model shows good balance between lightweighting and detection accuracy. Finally, the modified LAMP-pruned YOLOv11s-seg model is deployed on a Jetson Orin NX edge module and measured in a tea plantation, with the measured results exhibiting a detection speed of 34.1 FPS and verifying its availability in practical applications. Full article

Thin-film thermocouples (TFTCs) offer conformal sensing junctions with minimal thermal mass, enabling rapid transient response and direct deposition on curved or moving components, which are difficult to achieve using conventional wire thermocouples in applications such as high-speed machining, electric powertrain thermal management, and fuel-cell monitoring. In practical deployment, the effective accuracy of a TFTC can also be affected by the measurement setup used for calibration and testing, particularly lead-wire material transitions, cold-junction compensation, and wiring-related thermoelectric offsets. This study presents a systematic static calibration and performance evaluation of NiCr-based TFTCs under standardised laboratory conditions, with repeated measurements across the 20–260 °C range using both copper leads and matched compensation wires. The thermoelectric output exhibits excellent linearity; temperature reconstruction against a traceable standard reference yields a maximum deviation of approximately 0.27 °C, with root-mean-square and relative errors within tight bounds. Short-term extended-range verification up to 1000 °C confirms detectable thermoelectric signal generation under the present test conditions. A calibration data packet framework containing the calibrated TFTC sample, wiring configuration, calibration coefficients, validity range, and a GUM-compliant uncertainty budget is proposed to support consistent interpretation of calibration results in future digital integration. The study therefore provides a structured calibration workflow and uncertainty-reporting basis for the tested flexible NiCr-based TFTC configurations, supporting further reliability assessment, material-level characterisation, and digital integration. Full article