Existing surface defect detection methods for micro-motor commutators suffer from low detection accuracy, poor real-time performance, and high false detection and missed detection rates for small targets. To address these issues, this paper proposes a high-performance and robust commutator surface defect detection model (CLS-YOLO), using YOLOv11-n as the baseline model. First, a lightweight Cross-Scale Feature Fusion Module (CCFM) is introduced to integrate features from different scales, enhancing the model’s adaptability to scale variations and ability to detect small objects. This approach reduces model parameters and improves detection speed without compromising detection accuracy. Second, a Large Separable Kernel Attention (LSKA) module is incorporated into the detection head to strengthen feature understanding and capture, reducing interference from complex surface patterns on the commutator and significantly improving adaptability to various target types. Finally, to address issues related to the center point location, aspect ratio, angle, and sample imbalance in bounding boxes, SIoU Loss replaces the CIoU Loss in the original network, overcoming limitations of the original loss function and enhancing overall detection performance. Model performance was evaluated and compared on a commutator surface defect detection dataset, with additional experiments designed to verify the model’s effectiveness and feasibility. Experimental results show that, compared to YOLOv11-n, the CLS-YOLO model achieves a 2.08% improvement in mAP@0.5. This demonstrates that CLS-YOLO can accurately detect large defect targets while maintaining accuracy for tiny defects. Additionally, CLS-YOLO outperforms most YOLO-series models, including YOLOv8-n and YOLOv10-n. The model’s parameter count is only 1.860 million, lower than YOLOv11-n, with a detection speed increase of 8.34%, making it suitable for deployment on resource-limited terminal devices in complex industrial scenarios.

Deep learning has achieved significant success in the field of defect detection; however, challenges remain in detecting small-sized, densely packed parts under complex working conditions, including occlusion and unstable lighting conditions. This paper introduces YOLOv8-n as the core network to propose VEE-YOLO, a robust and high-performance defect detection model. Firstly, GSConv was introduced to enhance feature extraction in depthwise separable convolution and establish the VOVGSCSP module, emphasizing feature reusability for more effective feature engineering. Secondly, improvements were made to the model’s feature extraction quality by encoding inter-channel information using efficient multi-Scale attention to consider channel importance. Precise integration of spatial structural and channel information further enhanced the model’s overall feature extraction capability. Finally, EIoU Loss replaced CIoU Loss to address bounding box aspect ratio variability and sample imbalance challenges, significantly improving overall detection task performance. The algorithm’s performance was evaluated using a dataset to detect stranded elastic needle defects. The experimental results indicate that the enhanced VEE-YOLO model’s size decreased from 6.096 M to 5.486 M, while the detection speed increased from 179FPS to 244FPS, achieving a mAP of 0.926. Remarkable advancements across multiple metrics make it well-suited for deploying deep detection models in complex industrial environments.

In response to the current issues of poor real-time performance, high computational costs, and excessive memory usage of object detection algorithms based on deep convolutional neural networks in embedded devices, a method for improving deep convolutional neural networks based on model compression and knowledge distillation is proposed. Firstly, data augmentation is employed in the preprocessing stage to increase the diversity of training samples, thereby improving the model’s robustness and generalization capability. The K-means++ clustering algorithm generates candidate bounding boxes, adapting to defects of different sizes and selecting finer features earlier. Secondly, the cross stage partial (CSP) Darknet53 network and spatial pyramid pooling (SPP) module extract features from the input raw images, enhancing the accuracy of defect location detection and recognition in YOLO. Finally, the concept of model compression is integrated, utilizing scaling factors in the batch normalization (BN) layer, and introducing sparse factors to perform sparse training on the network. Channel pruning and layer pruning are applied to the sparse model, and post-processing methods using knowledge distillation are used to effectively reduce the model size and forward inference time while maintaining model accuracy. The improved model size decreases from 244 M to 4.19 M, the detection speed increases from 32.8 f/s to 68 f/s, and mAP reaches 97.41. Experimental results demonstrate that this method is conducive to deploying network models on embedded devices with limited GPU computing and storage resources. It can be applied in distributed service architectures for edge computing, providing new technological references for deploying deep learning models in the industrial sector.

This study enhances PEMFC performance at low temperature and low humidity by incorporating hydrophilic inorganic silica materials inside the PEMFC. Firstly, the polarization performance of the cell under varying humidity levels was investigated using a controlled variable method. Secondly, the power stability of the PEMFC was examined by observing voltage stability under constant current. Finally, the cell’s transient response at startup was studied. Experimental results indicate that the addition of silica partially mitigates the significant performance decline of the cell under a high humidity range (100% RH–50% RH). However, the effect is relatively average in a low humidity range (50% RH–20% RH).

Seal recognition, an essential basic technology for digitizing documents, plays a key role in modern document processing. Although optical character recognition (OCR) methods based on the Transformer architecture have made breakthroughs in several fields in recent years, the recognition task for the unique scenario of curved text still faces significant technical challenges. Existing solutions generally suffer from cumbersome processes and limited accuracy, which seriously restricts the applicability of practical applications. To address the above problems, this study innovatively combines the YOLO target detection framework with the TrOCR text recognition model to propose a solution for seal content recognition. The method significantly improves text recognition accuracy in complex scenes through three key aspects: precise localization, intelligent denoising, and efficient Recognition, and it demonstrates excellent bent text adaptation capability. Experimental results show that the proposed method achieves a high accuracy of 94.8% in the bent text recognition task. This research can widely apply to sealing instrument detection and Recognition.

Fuel cells, as one of the most promising alternatives to lithium-ion batteries for portable power systems, still face significant challenges. A critical issue is their substantial performance degradation under low-humidity conditions. To address this, researchers commonly add silica to components. This study employs a control variable method to systematically investigate the impact of four parameters—gas stoichiometry, temperature, humidity, and silica content—on fuel cell performance. Initially, the effects of gas stoichiometry, temperature, and humidity on performance were examined. Subsequently, hydrophilic silica was incorporated into the membrane electrode assembly (MEA) to assess its potential for improving performance in low-humidity environments. Experimental results reveal that under 100% humidification, silica addition had a minimal impact on performance, particularly at high temperatures where performance improved by only 2.5%. This is attributed to increased water production at elevated temperatures, which—when combined with silica’s water retention properties—exacerbates flooding. However, when humidity was reduced to 50%, silica incorporation significantly enhanced performance. At high temperatures, silica addition resulted in a 126.2% performance improvement, demonstrating its efficacy as a rational strategy under low-humidity conditions.