Hazardous Driving Scenario Identification with Limited Training Samples

Extracting hazardous driving scenarios from naturalistic driving data is essential for creating a comprehensive test scenario library for autonomous driving systems. However, due to the sporadic and low-probability nature of hazardous driving events, the available number of hazardous driving scenarios collected within a short period is limited, resulting in insufficient scenario coverage and a lack of diverse training samples. Such limited samples also lead to the poor generalization ability and weak robustness of the hazardous driving scenario identification model. To address these challenges, we propose a method to augment the limited driving scenario data through generative adversarial networks. By integrating DIG (Discriminator gradIent Gap) and APA (Adaptive Pseudo Augmentation) techniques into the original GAN framework, the quality of the generated data is enhanced. Benefiting from the advantages of augmented samples, we can leverage a more sophisticated ResNet architecture for feature extraction from compressed dashcam videos called motion profiles to identify hazardous driving scenarios. By incorporating the augmented samples into the training set, the AUC of the proposed hazardous driving scenario identification model is improved by 4% and surpassed the existing state-of-the-art methods.