Towards AI-powered personal shoppers of trendy clothes
In the past few years, there is a growing trend of using AI in the fashion field. On the one hand, the potential applications of generative AI have proved useful for creating product descriptions on e-commerce or supporting fashion designers to create new models based on data provided by customers. On the other hand, customers could exploit AI to compose personalized fashioned outfits with compatible garments by effective dialoguing.
In this regard, early works have mainly focused on recommendations by scoring visual appeal and representing garments as ordered sequences or as collections of pairwise-compatible items. However, suggesting complementary clothing items to compose an outfit involves a fine understanding of both fashion trends and visual aesthetics. The work carried out in this paper, tries to build a bridge between outfit recommendation and generation, by discovering new appealing outfits starting from a collection of pre-existing ones. We propose a transformer-based architecture that exploits multi-headed self-attention to capture relations between clothing items in a graph as a message-passing step in Convolutional Graph Neural Networks. We will also discuss preliminary experiments to create reactive digital signage that learns to optimize the recommendations by iteratively adjusting machine proposals according to the user’s feedback with a level of professionalism remotely close to human intelligence.
Prof. Del Bimbo is a Professor at the Department of Information Engineering of the University of Firenze. He is the author of over 350 scientific publications in computer vision, multimedia content analysis, indexing, and retrieval. He was the General Chair of ACM Multimedia 2022, ICPR 2020, ECCV 2012, ICMR 2011, ACM Multimedia 2010, and IEEE ICMCS 1999, and the Program Chair of ICPR 2016, and ICPR 2012, and ACM Multimedia 2008. He was the Editor in Chief of ACM TOMM Trans. on Multimedia Computing, Communications, and Applications and Associate Editor of IEEE Trans. on Pattern Analysis and Machine Intelligence, IEEE Trans. on Multimedia, Pattern Recognition, Multimedia Tools and Applications, and Pattern Analysis and Applications. Prof. Del Bimbo is an IAPR Fellow and the recipient of the 2016 ACM SIGMM Award for Outstanding Technical Contributions to Multimedia Computing, Communications, and Applications. He is presently the Chair of ACM SIGMM the ACM Special Interest Group in Multimedia.
Research activities at INA ( French National Audiovisual Archive)
In this presentation, we’ll be looking at the various research activities carried out at Ina, and the opportunities they open up, both in terms of industrialization within the information system, and in terms of enhancing the value of our collections. This mainly concerns the fields of multimedia content analysis (video, sound and text), indexing and preservation. We will also outline the potential directions for our future work, including the impact of deep learning and the importance of transdisciplinary work in our approaches and collaborations.
Nicolas Hervé, Head Researcher
Nicolas Hervé is the Head of the Research Department at Ina. His current research activities focus primarily on the digital social sciences and humanities and, more specifically, on the information ecosystem using machine learning and data mining approaches. He coordinates the OTMedia research platform that enables transdisciplinary studies on the media world and the propagation of online information.