Research on Small Object Detection in Intelligent Driving Based on Data Mining
Small object detection is an important research topic in the field of intelligent driving safety,espe-cially in the detection of traffic signs,which is a key link in the perception of intelligent driving environ-ments.However,due to the long-distance recognition of traffic signs and the small size of the target,prob-lems such as missed detection or low recognition accuracy often occur.Based on the Faster R-CNN algo-rithm,the backbone network VGG16 is replaced with ResNet50,a hybrid attention mechanism is integrated into the backbone residual structure,and a muti-scale sliding windows is used to improve the RPN net-work.Feature maps are generated and fused at different depth convolution layers.The improved algorithm has increased the detection accuracy mAP from 85.99％ to 94.38％,effectively improving the ability to recognize small targets of traffic signs in intelligent driving scenarios.
万方数据
维普
