Vehicle danger test scene identification method based on adaptive screening

The invention discloses a vehicle danger test scene identification method based on adaptive screening, and relates to the technical field of vehicle safety. According to the method, two different classifier models are used at the same time for training, additional training scene parameters are introduced to predict the two classifier models based on the training process of an original training scene set, and then the prediction result of the two classifier models is obtained according to the training scene parameters with different prediction results. Training scene sets corresponding to the classifier models are continuously updated for adaptive screening training, gradual adaptive optimization of the two classifier models is achieved, and the technical effect of improving the recognition precision of the vehicle danger test scene recognition model is achieved. 本发明公开了一种基于自适应筛选的车辆危险测试场景识别方法，涉及车辆安全技术领域。该方法通过同时使用两种不同的分类器模型进行训练，并基于原有训练场景集的训练过程基础，引入额外的训练场景参数分别进行两个分类器模型的预测，再根据二者预测结果相异的训练场景参数，不断更新分类器模型各自对应的训练场景集进行自