Where to Buy It: Matching Street Clothing Photos in Online Shops

Spatio-Temporal Wardrobe Generation of Actors’ Clothing in Video Content

“…Results on the region of interest based classification task results in a top-1 accuracy of 15% and a top-3 accuracy of 49%. The accuracy of the clothing type classification task is relatively low, but similar results are obtained in state-of-the-art approaches [19,20]. This accuracy shows the necessity for manual verification of the classification results.…”

“…In recent years, the research community has actively been focusing on the automatic classification and detection of clothes in digital content [19][20][21]. Unfortunately, due to the large visual differences between images on e-commerce websites (photographed on a clean, white background with clear lighting conditions) and those retrieved from in-the-wild videos, this problem is still largely unsolved.…”

Multi-modal Embedding for Main Product Detection in Fashion

“…In the case of fashion e-commerce, a specific item being sold is normally depicted worn by a model and tastefully combined with other garments to make it look more attractive. Existing approaches for recommendation or retrieval focus on images only, and normally require hard-to-obtain datasets for training [7], omitting the metadata associated with the ecommerce products such as titles, colors, series of tags, descriptions, etc. that can be used to improve the information obtained from the images.…”

Regularized deep learning for fashion and social media data analysis

“…Most existing on-line shopping websites only provide the images of the product itself, or better, provide how skinny model wearing the clothing with clean background. Cross-domain retrieval and recommendation remains a challenging problem due to the large discrepancy between different domains, such as the background, pose, illumination [12,13,14,15].…”

“…They are build based on the bi-directional, (i.e., both street-to-shop and shop-to-street), cross-domain clothing retrieval. In street-to-shop clothing retrieval [11,12], given a daily human photo taken in the street view, similar garments from on-line shops are retrieved using the proposed cross-domain image retrieval solution. In shop-to-street clothing retrieval, for a shop item (e.g., a blue skirt), we show how ordinary people wearing this item on street by shop-to-street clothing retrieval.…”