Abstract

Recommendation in the fashion domain has seen a recent surge in research in various areas, for example, shop-the-look, context- aware outfit creation, personalizing outfit creation, etc. Majority of state of the art in the domain of outfit recommendation pursue to improve compatibility among items so as to produce high quality outfits. Some recent works have realized that style is an important factor in fashion and have incorporated it in compatibility learning and outfit generation. These methods often depend on the availability of fine-grained product categories or the presence of rich item attributes (e.g., long-skirt, mini- skirt, etc.). In this work, we aim to generate outfits conditional on styles or themes as one would dress in real life, operating under the practical assumption that each item is mapped to a high level category as driven by the taxonomy of an online portal, like outdoor, formal etc and an image. We use a novel style encoder network that renders outfit styles in a smooth latent space. We present an extensive analysis of different aspects of our method and demonstrate its superiority over existing state of the art baselines through rigorous experiments.

Read more in the paper.
In ECIR 2022 full paper track (presentation). Initial version presented at CODS COMAD 2022 Demo track.

Illustration of the effectiveness of style-guided outfit generation over a style independent variant. Given a top-wear liked by a user, a style-independent compatibility model will accept only outfits which have the dominant style while style-guided methods will accept outfits from multiple styles as well as reject the outfit which is pre-conditioned with a wrong style

Paper and Supplementary Material

ECIR


@inproceedings{style_recommendation22,
  title={Recommendation of Compatible Outfits Conditioned on Style},
  author={Banerjee, Debopriyo and Dhakad, Lucky and Maheshwari, Harsh and Chelliah, Muthusamy and Ganguly, Niloy and Bhattacharya, Arnab},
  booktitle={European Conference on Information Retrieval},
  pages={35--50},
  year={2022},
  organization={Springer}
}

CODS COMAD


@inproceedings{banerjee2022application,
              title={An Application to Generate Style Guided Compatible Outfit},
              author={Banerjee, Debopriyo and Maheshwari, Harsh and Dhakad, Lucky and Bhattacharya, Arnab and Ganguly, Niloy and Chelliah, Muthusamy and Agarwal, Suyash},
              booktitle={5th Joint International Conference on Data Science \& Management of Data (9th ACM IKDD CODS and 27th COMAD)},
              pages={260--264},
              year={2022}
            }

Qualitative Results

Style conditioning enables us to create outfits for different styles given the same parent item. Thanks to the smooth style latent space, we can linearly traverse within the latent space from one style to the other and create outfits which are a combination of such styles.