and difficulties in subsequent processing of recognition results in the manual identification of secondary substation diagrams
a cross-scale visual recognition method for substations based on an improved YOLOv10 is proposed. To cope with the diversity of target scales and scene complexity
a multi-scale dilated residual structure is adopted to expand the receptive field and enhance the model's adaptability to multi-scale symbols. A Balanced Detail Fusion (BDF) module is designed to promote the adaptive fusion of shallow detail features and deep semantic features. The DySample dynamic upsampling mechanism is introduced to replace traditional interpolation methods
effectively improving the resolution recovery quality and edge alignment accuracy of feature maps. Furthermore
the Focaler-SIoU loss function is proposed
which introduces directional and scale constraints to enhance the model's boundary regression capability for small targets and occluded symbols. Experimental results show that this method significantly improves detection performance for densely arranged
small-sized
and morphologically complex targets while maintaining high inference efficiency.
BUMBÁLEK R,UFITIKIREZI M D D J,UMURUNGI N S,et al. Computer vision in precision livestock farming: benchmarking YOLOv9,YOLOv10,YOLOv11,and YOLOv12 for individual cattle identification [J ] . Smart Agricultural Technology,2025,12: 101208-101219.
ZHANG C,LIU X,SHANG Q,et al. Personnel target detection in infrared environment based on YOLOv3-tinier network and its FPGA implementation [J ] . Infrared Physics and Technology,2025,150: 106015-106027.
RONG X,ZHONG C. PPFGED: Federated learning for graphic element detection with privacy preservation in multi-source substation drawings [J ] . Expert Systems with Applications,2024,243: 122758-122770.
BHANBHRO H,HOOI K Y,KUSAKUNNIRAN W,et al. A symbol recognition system for single-line diagrams developed using a deep-learning approach [J ] . Applied Sciences,2023,13(15): 182-193.
FAHIM F,HASAN S M. Enhancing the reliability of power grids: A YOLO based approach for insulator defect detection [J ] . e-Prime-Advances in Electrical Engineering,Electronics and Energy,2024,9: 100663-100671.
ZHANG Y,CHEN X,SUN S,et al. Vehicle detection in drone aerial views based on lightweight OSD-YOLOv10 [J ] . Scientific Reports,2025,15(1): 25155-25163.
ZHAO F,XU D,REN Z,et al. Mamba-based super-resolution and semi-supervised YOLOv10 for freshwater mussel detection using acoustic video camera: A case study at Lake Izunuma,Japan [J ] . Ecological Informatics,2025,90: 103324-103332.