The Comprehensive Cars (CompCars) dataset contains data from two scenarios,including images from web-nature and surveillance-nature. The web-nature datacontains 163 car makes with 1,716 car models. There are a total of 136,726images capturing the entire cars and 27,618 images capturing the car parts.The full car images are labeled with bounding boxes and viewpoints. Each carmodel is labeled with five attributes, including maximum speed, displacement,number of doors, number of seats, and type of car. The surveillance-naturedata contains 50,000 car images captured in the front view.The dataset is well prepared for the following computer vision tasks:Fine-grained classificationAttribute predictionCar model verificationThe train/test subsets of these tasks introduced in our paper are included inthe dataset. Researchers are also welcome to utilize it for any other taskssuch as image ranking, multi-task learning, and 3D reconstruction.Related publications:Linjie Yang, Ping Luo, Chen Change Loy, Xiaoou Tang. A Large-Scale CarDataset for Fine-Grained Categorization and Verification, In Computer Visionand Pattern Recognition (CVPR), 2015. (PDF)