Accurate detection of steel surface defects is of great significance for ensuring product quality and production safety. However, existing detection models still suffer from insufficient accuracy and poor robustness when facing practical industrial scenarios characterized by complex defect morphologies, diverse scales, blurred boundaries, and strong background interference. To address this, this paper proposes an improved YOLO11 detection model – DSP-YOLO, aimed at enhancing the comprehensive per-formance of steel surface defect detection. We design a Dynamic Attention-mixed Normalized Block, which introduces multi-scale direction-aware dynamic depthwise convolution and channel interaction mechanisms to strengthen the flexibility and ex-pressive power of feature extraction; propose a Semantic Context Fusion Module that integrates semantic awareness and context-guided strategies to effectively improve feature fusion effectiveness under complex textured backgrounds; additionally, intro-duce a lightweight asymmetric pinwheel-shaped convolution structure to significantly enhance the receptive field and directional modeling capability while maintaining low computational cost. Experimental results on two typical steel defect datasets, NEU-DET and GC10-DET, show that the proposed method achieved 2.5% and 2.6% improvements in mAP@0.5 respectively, surpassing existing mainstream YOLO models in terms of accuracy, efficiency, and generalization ability, demonstrating good potential for in-dustrial applications.

In order to understand the current status of the application of NDT technology in storage tanks and pressure vessels during the past two decades, this study quantitatively analyzed and visualized the literature in the Web of Science database for the period from January 2004 to July 2024 using VOSviewer and CiteSpace. The hotspots and evolutionary paths of NDT technology applications in storage tanks and pressure vessels were detailed by descriptive statistics, co-occurrence analysis, burst word analysis and co-citation analysis. Meanwhile, the major published journals, highly cited literature and highly cited journals were analyzed in depth. The results show that (1) Chinese scholars have contributed the most papers in this field; (2) a stable collaborative group has been formed in this field; (3) acoustic emission technology is the most popular NDT technology; and (4) the NDT technology is gradually maturing and is widely used in industrial practice.

In cotton leaf disease detection, low-light conditions can cause the loss of fine-grained lesion details, while lesions are often small and irregular in shape; these factors jointly aggravate missed detections and false detections. To address these issues, we propose a novel cotton leaf disease detection model, ADD-YOLO. A cotton leaf disease dataset was constructed, and YOLO11n was adopted as the baseline network. First, an Adaptive Lighten Cross-Attention (ALCA) lightweight cross-attention module was designed to enhance lesion recognition under low-light conditions. Second, a Deformable Attention Graph Enhancement Module (DAGEM) was introduced to strengthen feature representation for small lesions. Finally, Depthwise Separable Convolution (DWConv) was incorporated into the backbone to reduce computational overhead. Experimental results show that, compared with the baseline model, ADD-YOLO improves Precision, Recall, mAP@0.5, and mAP@[0.5:0.95] by 6.0%, 3.1%, 4.4%, and 5.2%, respectively, while reducing FLOPs and parameter counts by 9.5% and 15.9%, respectively. The model achieves 198.6 FPS, meeting real-time requirements for field deployment. Moreover, generalization experiments on public datasets demonstrate the robustness and transferability of the proposed method, indicating its effectiveness for cotton leaf disease detection tasks.

To understand the recent advancements in the application of digital twin technology in the field of robotics, this study utilizes VOSviewer and CiteSpace to conduct a quantitative analysis and visualization of relevant literature from the Web of Science database from 2019 to June 2024. Through descriptive statistics, co-occurrence analysis, timeline analysis, and co-citation analysis, the study presents the research hotspots and evolutionary paths of global digital twin applications in the field of robotics. Additionally, it analyzes the main publishing journals, highly cited papers, and highly cited journals. The findings reveal that: (1) Chinese scholars have contributed the most papers in this field, while the research of British scholars has the highest recognition; (2) stable research themes have formed in this field, including robotic system architecture, human-robot collaboration, and virtual simulation; (3) the research focus of digital twin technology in the field of robotics has shifted from macro-system optimization to micro-technical details and advanced applications.