1 results listed
This paper introduces a recommendation system based on deep learning that is intended to recommend
accessories after users upload images of t-shirts, pants or even sarees. Users are able to upload any of these items and the
system will recommend accessories specific to each item. Fashion recommendations have always been a hot topic, this paper
solves one of the deep problems of fashion recommendations. It employs a multiclass classification method in the first stage
that involves a general classifier which first evaluates the clothing type and then evaluates the subcategory gen, colour and
design. For example, a T-shirt would be broken down further into gender-specific sleeves, number of sleeves, colours, and
different designs. These attributes are then indexed into a recommendation structure to retrieve suitable accessories such pants,
shoes, watches, sunglasses, bangles, and rings. The user interface is created in React to optimize the user experience while
Flask is used as a backend for REST APIs. This project utilizes a modular system as well as advanced Deep Learning
architectures and guarantees improvements over existing solutions for fashion recommendations through machine learning by
fostering greater personalization and precision of recommendations. A prospective enhancement may include monitoring user
preferences and providing feedback to maximize the system’s value for e- commerce databases.
International Conference on Advanced Technologies, Computer Engineering and Science
ICATCES
Tulsi Choudhari
Vivek Madhavi
Nikita Ghadge
Mansi Deore