Wood surface defect detection is a critical step in wood processing and manufacturing. To address the performance degradation caused by small targets and multi-scale features in wood surface defect detection, a novel deep learning model is proposed in this study, FDD-YOLO, specifically designed for this task. In the feature extraction stage, the C2f module and the funnel attention (FA) mechanisms are integrated into the design of the C2f-FA module to enhance the model’s ability to extract features of wood surface defects of various sizes. Additionally, the Dual Spatial Pyramid Pooling-Fast (DSPPF) module is developed, and the Context Self-attention Module (CSAM) is introduced to address the limitations of traditional max pooling methods, which often overlook global contextual information when extracting local features, thereby improving the detection of small-scale wood defects. In the feature fusion stage, a Dual Cross-scale Weighted Feature-fusion (DCWF) module is proposed to fuse shallow, deep, and cross-scale features through a weighted summation approach, effectively addressing the challenge of scale variation in wood surface defects. Experimental results demonstrate that the proposed FDD-YOLO model significantly improves detection performance, increasing the mAP of the baseline model YOLOv8 from 78% to 82.3%, a substantial enhancement of 4.3 percentage points. Furthermore, FDD-YOLO outperforms other mainstream defect detection models in terms of detection accuracy. The proposed model demonstrates significant potential for industrial applications by improving detection accuracy, enhancing production efficiency, and reducing material waste, thereby advancing quality control in wood processing and manufacturing.

Wood plate bark removal processing is critical for ensuring the quality of wood processing and its products. To address the issue of lack of datasets available for the application of deep learning methods to this field, and to fill the research gap of deep learning methods in the application field of wood plate bark removal equipment, a benchmark for wood plate segmentation in bark removal processing is proposed in this study. Firstly, a costumed image acquisition device is designed and assembled on bark removal equipment to capture wood plate images in real industrial settings. After data filtering, enhancement, annotation, recording, and partitioning, a benchmark dataset named the WPS-dataset containing 4863 images was constructed. The WPS-dataset is evaluated by training six typical semantic segmentation models. The experimental results show that the models effectively learn and understand the WPS-dataset characteristics during training, resulting in high performance and accuracy in wood plate segmentation tasks. The WPS-dataset can lay a solid foundation for future research in bark removal processing and contribute to advancements in this field.

In order to accurately detect the temperature of molten aluminum and overcome the adverse influence of high temperature and corrosiveness on the sensing results, a temperature detection system based on a multi-node sapphire fiber sensor was proposed and developed. Through the structural parameter design of the fiber sensor, the scheme of utilizing the 0.7 mm diameter fiber and 0.5 mm groove was formulated. Simulation and analysis were carried out to determine the ultrasonic response distribution of the signal passing through the whole fiber sensor. The results indicate that the system is capable of distinguishing test signals from various positions and temperatures. Following the completion of the static calibration, the temperature of the molten aluminum was observed in real-time, and the data of the temperature measurements conducted at the two groove locations were compared. According to the obtained results, the test accuracy was greater than 1 degree Celsius and the temperature test stability was good, laying a solid foundation for the potential development of temperature measurement devices.

Addressing the problem of photon multiple scattering interference caused by turbid media in optical measurements, biomedical imaging, environmental monitoring and other fields, existing Monte Carlo light scattering simulations widely adopt the Henyey-Greenstein (H-G) phase function approximation model. However, traditional computational resource limitations and high numerical complexity have constrained the application of precise scattering models. Moreover, the single-parameter anisotropy factor assumption neglects higher-order scattering effects and backscattering intensity, failing to accurately characterize the multi-order scattering properties of complex media. To address these issues, we propose a GPU-accelerated Monte Carlo-Rigorous Mie scattering transport model for complex scattering environments. The model employs rigorous Mie scattering theory to replace the H-G approximation, achieving efficient parallel processing of phase function sampling and complex scattering processes through pre-computed cumulative distribution function optimization and deep integration with CUDA parallel architecture. To validate the model accuracy, a standard scattering experimental platform based on 5μm polystyrene microspheres was established, with multiple optical depth experimental conditions designed, and spatial registration techniques employed to achieve precise alignment between simulation and experimental images. The research results quantitatively demonstrate the systematic accuracy advantages of rigorous Mie scattering phase functions over H-G approximation in simulating lateral scattering light intensity distributions, providing reliable theoretical foundations and technical support for high-precision optical applications in complex scattering environments.

Detecting small unauthorized UAVs against complex backgrounds is challenging: targets can be just a few pixels wide, background clutter is pervasive, and the detection system must run on resource-constrained edge hardware. This paper presents a lightweight detector built on YOLOv11 that jointly addresses background suppression and edge-deployment efficiency. A serial channel-spatial attention module is embedded into the backbone to sharpen target features and suppress interference. DySample preserves edge details during resolution recovery, and the detection head is restructured by adding a high-resolution P2 branch while removing the large-object P5 head, shifting the focus to small targets without inflating the model. The WIoU v3 loss function stabilizes training by automatically down-weighting low-quality samples. On the DUT Anti-UAV dataset, the proposed model obtains 92.2% mAP@0.5 and 61.5% mAP@0.5:0.95 with only 2.11 M parameters, a reduction of 18.2% relative to the YOLOv11n baseline. Its ability to generalize across different scenarios is verified by cross-dataset evaluation on Det-Fly and LRDDv2. Deployed on an NVIDIA Jetson Orin Nano Super with TensorRT FP16 optimization, the system achieves 37 end-to-end FPS. These findings demonstrate that the presented method significantly improves small-target detection in complex backgrounds while maintaining a lightweight design suitable for real-time edge deployment.