Introduction
There is more to images than their objective physical content: for example, advertisements are created to persuade a viewer to take a certain action. We propose the novel problem of automatic advertisement understanding. To enable research on this problem, we create two datasets: an image dataset of 64,832 image ads, and a video dataset of 3,477 ads. Our data contains rich annotations encompassing the topic and sentiment of the ads, questions and answers describing what actions the viewer is prompted to take and the reasoning that the ad presents to persuade the viewer ("What should I do according to this ad, and why should I do it?"), and symbolic references ads make (e.g. a dove symbolizes peace). We also analyze the most common persuasive strategies ads use, and the capabilities that computer vision systems should have to understand these strategies. We present baseline classification results for several prediction tasks, including automatically answering questions about the messages of the ads.
[top]Publications
- Cross-Modality Personalization for Retrieval. Nils Murrugarra-Llerena and Adriana Kovashka. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2019. (Oral.) [pdf] [supp] [poster] [data]
- ADVISE: Symbolism and External Knowledge for Decoding Advertisements. Keren Ye and Adriana Kovashka. To appear, Proceedings of the European Conference on Computer Vision (ECCV), September 2018. [pdf] [supp] [related code]
- Story Understanding in Video Advertisements. Keren Ye, Kyle Buettner, Adriana Kovashka. To appear, Proceedings of the British Machine Vision Conference (BMVC), September 2018. [pdf]
- Persuasive Faces: Generating Faces in Advertisements. Christopher Thomas and Adriana Kovashka. To appear, Proceedings of the British Machine Vision Conference (BMVC), September 2018. [pdf] [supp]
- Automatic Understanding of Image and Video Advertisements. Zaeem Hussain, Mingda Zhang, Xiaozhong Zhang, Keren Ye, Christopher Thomas, Zuha Agha, Nathan Ong, Adriana Kovashka. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), July 2017. (Spotlight) [pdf] [supplementary] [poster] [spotlight]
[top]
Funding
NSF CISE CRII: RI: Automatically Understanding the Messages and Goals of Visual Media (Award #1566270)
Google Faculty Research Awards
This material is based upon work supported by the National Science Foundation under Grant Number 1566270. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.
[top]Image Data and Annotations
[interface with new object transformation annotations]Explore the image dataset visualization here
(thanks to Mingda Zhang and Narges Honarvar Nazari)
[original interface] Explore the image dataset visualization here
Type | Count | Example |
Topic | 204,340 | Electronics |
Sentiment | 102,340 | Cheerful |
Q+A | 202,090 | I should bike because it’s healthy. |
Symbol | 64,131 | Danger (+ bounding box) |
Strategy | 20,000 | Contrast |
Slogan | 11,130 | Save the planet... save you. |
- Readme
- Download images: Email us for image URLs
- Download image annotations
- Ad or not-ad classifier (thanks to Chris Thomas)
Video Data and Annotations
Type | Count | Example |
Topic | 17,345 | Cars and automobiles, Safety |
Sentiment | 17,345 | Cheerful, Amazed |
Action/Reason | 17,345 | I should buy this car because it is pet-friendly. |
Funny? | 17,374 | Yes/No |
Exciting? | 17,374 | Yes/No |
English? | 15,380 | Yes/No/Does not matter |
Effective? | 16,721 | Not/.../Extremely Effective |
Contact
For any questions, issues, concerns, and comments, please email Adriana Kovashka at kovashka AT cs DOT pitt DOT edu[top]