Design of traffic event detection system based on deep learning
The establishment of comprehensive traffic incident detection systems have become an important component of China's intelligent transportation system.This paper analyzes the characteristics of traffic event detection scenes in depth from both data and algorithms,and proposes a traffic event detection system based on deep learning.A hybrid architecture joint learning network is introduced,addressing the challenges of multi-label classification in image data by comprehensively leveraging the advantages of ViT and Swin Transformer.A series of data augmentation methods have been designed to cope with the impact of data imbalance on deep learning models,and effectively alleviating the problem of model overfitting.The experimental results demonstrate that the system has better accuracy and generalization ability in traffic event detection.The system has been applied to multiple practical projects,and has achieved favorable application outcomes.
万方数据
维普
