My research interests lie broadly in artificial intelligence and machine learning, including the applications in computer vision and natural language processing. I am very lucky to have Prof. Adriana Kovashka and Prof. Rebecca Hwa as my co-advisors.
The best way to reach me is via my email: firstname.lastname@example.org.
(2018.9) I attended BMVC 2018 at Newcastle upon Tyne, UK and gave a spotlight talk for our paper.
(2018.9) The dataset for our BMVC 2018 spotlight paper was released in the project page.
(2016.11) I received the Arts and Sciences Graduate Fellowship again for 2017 Spring.
(2016.5) I received the Arts and Sciences Graduate Fellowship for 2016 Fall.
Automatic Advertisement Understanding
We propose a novel problem of automatic advertisement understanding, and create two new datasets containing rich annotations to facilitate relevant research. We also analyze the most common persuasion 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 ads.
Z. Hussain, M. Zhang, X. Zhang, K. Ye, C. Thomas, Z. Agha, N. Ong, A. Kovashka.
Automatic Understanding of Image and Video Advertisements. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), July 2017.(Spotlight)
Implicit Relationship between Images and Text in Ads
Images and text in advertisements interact in complex, non-literal ways. The two channels are usually complementary, with each channel telling a different part of the story. Current approaches, such as image captioning methods, only examine literal, redundant relationships, where image and text show exactly the same content. To understand more complex relationships, we first collect a dataset of advertisement interpretations for whether the image and slogan in the same visual advertisement form a parallel or non-parallel relationship, with the help of workers recruited on Amazon MTurk. We develop a variety of features that capture the creativity of images and the specificity or ambiguity of text, as well as methods that analyze the semantics within and across channels. We show that our method outperforms standard image-text alignment approaches on predicting the parallel/non-parallel relationship between image and text.
Mingda Zhang, Rebecca Hwa, Adriana Kovashka.
Equal But Not The Same: Understanding the Implicit Relationship Between Persuasive Images and Text. Prceedings of the British Machine Vision Conference(BMVC), September 2018.(Spotlight)
Automatic Understanding of Image and Video Advertisements.(Spotlight)
2017 Computer Vision and Pattern Recognition (CVPR), Honolulu, HI.
Equal But Not The Same: Understanding the Implicit Relationship Between Persuasive Images and Text.(Spotlight)
2018 British Machine Vision Conference (BMVC), Newcastle upon Tyne, United Kingdom.
(2018.5 - 2018.8) Ph.D. Software Engineering Intern at Google AI (Research & Machine Intelligence), Seattle, WA