DVM-Car

A Large-Scale Dataset for Automotive Applications

This publicly available dataset aims to facilitate business related research and applications in automotive industry such as car appearance design, consumer analytics and sales modelling. To use, please go through the user manual [PDF].

Download

Part I. Sale and Specification Data

  • Basic table: car attributes such as model name, model ID and brand name.
  • Sales table: ten years car sales data in UK/GB.
  • Price table: entry-level (i.e. the cheapest trim price) new car prices across years.
  • Trim table: trim attributes like the selling price (trim level), engine type and engine size.
  • Ad table: more than 0.25 million used car advertisements.
  • Image table: car images attributes like colour and viewpoint.

    • Download: Sales and specification tables (CSV)

    Part II. Image Data

  • 1,451,784 images from 899 UK market car models.
  • Cover models over the last two decades.
  • All resized to 300x300 resulutions with background removed.
  • Predicted viewpoints and quality check ressults are in image table.
  • Store under the structure "Brand-Model-Year-Colour".

    • Download: Car images (13.6 GB)
    Download: Quality checked front-view images (730 MB)

    Structure

    Timeline

    2018

    June 2018

    A survey study is conducted with business researchers and computer scientists to explore the most common issues which they face when using the existing datasets.

    2018

    September 2018

    Data collection and preparation start. Car image and non-visual data from various sources are collected and integrated. More than six million raw images have been gathered.

    2019

    August 2019

    DVM-CAR 1.0 is released via the GitHub hosted website https://deepvisualmarket-ing.github.io.

    2020

    May 2021

    DVM-CAR 2.0 is released! Now all the images are resized to 300x300 resolutions; segment results are no longer provided directly; Image data of 2019 registered car models is added and the non-visual feature data is updated to 2020.

    2022

    Nov 2022

    The corresponding paper, "DVM-CAR: A large-scale automotive dataset for visual marketing research and applications," is now accepted by IEEE BigData 2022.

    Citation

    • Important: Researcher shall use this dataset only for non-commercial research and educational purposes.
    • If you find the dataset useful with your research, you can use the following citation:

    Jingming Huang, Bowei Chen, Lan Luo, Shigang Yue, and Iadh Ounis. (2022). "DVM-CAR: A large-scale automotive dataset for visual marketing research and applications". In Proceedings of IEEE International Conference on Big Data, pp.4130–4137. [PDF_link]