To ensure the safety of live working operaters

it proposes a live working area prediction algorithm based on the fusion of visible light and infrared imaging and the improved you only look once version 8(YOLOv8) algorithm model.It improves the accuracy and reliability of the live working boundary through the multimodal data complementarity method.Firstly

among these improvements

it introduces the simple parameter-free attention module(SimAM)attention mechanism into YOLOv8 to enhance feature extraction capabilities.Secondly

it incorporates adaptive deformable convolution adaptive deformable convolution v2(A-DCNv2)to improve the detection accuracy of small targets.Finally

it fuses the visible light detection results with infrared temperature data using an adaptive weighted fusion algorithm to optimize boundary determination

ensuring precise positional matching between high/low-temperature areas and visual targets.Experimental results demonstrate that the proposed algorithm enhances the model's robustness for target detection with almost no increase in computational load.