Datasets

The Car Frontal Styling Database – CFSDB

News

2017-05-19 First release of CFSDB.

Description

CFSDB mainly focuses on the data-driven car frontal styling analysis task, thus contains the car images form the frontal view with the styling ROI, i.e. the union area of headlights, the grille, and the bumper. Because of the different aspect ratios of styling ROI between sedan and SUV, two datasets are built accordingly. The sedan dataset contains 4726 models corresponding to 22 car brands, and the SUV dataset contains 2441 models corresponding to 21 car brands. The selected car brands are high ranking in the sales of the Chinese automobile market, and all the car models collected are mainly produced after 2005. All the car images are downloaded from the Internet and are manually processed, including the styling ROI extraction, background removal and car plate removal. In each dataset, the car images are classified into different folders according to their brands.

The CFSDB is well prepared for car styling analysis and other fine-grained categorization tasks.

Note

1. You need to complete the release agreement form to download the dataset. Please see below.

2. The CFSDB database is available for non-commercial research purposes only.

3. All images of the CFSDB database are obtained from the Internet which are not property of DUT, Dalian University of Technology. The DUT is not responsible for the content nor the meaning of these images.

4. You agree not to reproduce, duplicate, copy, sell, trade, resell or exploit for any commercial purposes, any portion of the images and any portion of derived data.

5. You agree not to further copy, publish or distribute any portion of the CFSDB  database. Except, for internal use at a single site within the same organization it is allowed to make copies of the database.

6. The AutoMorpher reserves the right to terminate your access to the database at any time.

Download instructions

Download the CFSDB dataset Release Agreement, read it carefully, and complete it appropriately. Note that the agreement should be signed by a full-time staff member (that is, student is not acceptable). Then, please scan the signed agreement and send it to Prof. Baojun Li (bjli(at)dlut.edu.cn). We will verify your request and contact you on how to download the database.