Motorcycle and Electric Vehicle Traffic Violation Recognition Dataset

The current transportation industry faces increasingly serious issues of traffic violations, particularly with higher violation rates among motorcycles and electric vehicles, leading to frequent accidents. Existing monitoring systems largely rely on manual review, which is inefficient and prone to errors, highlighting the urgent need for an efficient automated recognition solution. This dataset aims to provide high-quality image data of traffic violations by motorcycles and electric vehicles to support the training of deep learning models for automated violation recognition. Data collection was conducted using high-resolution cameras in urban roads and busy traffic areas to ensure the diversity and authenticity of the data. We employed multiple rounds of labeling and expert review quality control measures to ensure annotation accuracy and consistency. Data is stored in JPG format and label information is organized through JSON files to facilitate subsequent model training and testing. This dataset not only achieves annotation accuracy of over 95% but also improves model robustness through data augmentation techniques such as random cropping and rotation, significantly enhancing recognition accuracy. Initial tests indicate a 20% improvement in recognition rates.