Highway Severe Accident Detection Dataset

In the current transportation industry, frequent traffic accidents lead to significant economic losses and casualties, especially on highways where the speed of accident detection and response is crucial. However, existing accident detection systems commonly rely on manual monitoring, which is inefficient and prone to missed reports. To address this issue, we have constructed the Highway Severe Accident Detection Dataset to provide high-quality training data for intelligent transportation systems, improving the accuracy and timeliness of automatic detection technology. This dataset is collected using high-resolution cameras on highways and includes images before and after accidents, ensuring diversity and authenticity. We have employed multi-round annotations and expert reviews to ensure high annotation accuracy. A total of 5000 images have been collected, stored in JPG format, with a file size of approximately 1.2G, organized so that each image has a corresponding metadata file for easy subsequent processing and analysis.