High-Speed Rear-End Vehicle Damage Segmentation Dataset

The current transportation industry faces increasingly severe safety challenges, especially with the frequent occurrence of rear-end collisions on highways, causing casualties and property damage. Existing accident analysis methods mostly rely on manual observation, which is inefficient and prone to errors. To address this issue, this dataset focuses on the damage segmentation of high-speed rear-end vehicles by providing high-quality images and detailed mask annotations, aiming to improve the accuracy and efficiency of automated accident analysis. The construction of the dataset includes extracting images from real traffic monitoring videos, with annotators ensuring data accuracy through multiple rounds of reviews. The data is stored using JPG format images and JSON format annotation files to facilitate subsequent deep learning model training and testing.