Method and device for testing robustness of vehicle detection model based on deep learning

The invention discloses a method and a device for testing the robustness of a vehicle detection model based on deep learning. The method comprises the following steps: acquiring a vehicle picture and a corresponding CAD (Computer Aided Design) model; establishing and training a parameterized texture generation network model, performing feature extraction on a vehicle picture through a deep neural network to obtain a first feature, performing feature extraction on a CAD model through a triangular mesh model to obtain a second feature, and connecting the first feature and the second feature to serve as input of a deep convolutional generation network, and taking a parameterized texture picture as the output of the deep convolutional generation network, processing the parameterized texture picture to obtain a synthesized vehicle picture, inputting the synthesized vehicle picture into a vehicle classification network, recording the parameterized texture picture classified as a non-vehicle picture in a judgment re