Cognitive analysis and classification of apparel images

Abstract

The present invention relates to cognitive analysis and classification of apparel images. The present disclosure relates to convolutional neural networks, and more particularly, to performing cognitive analysis of images of apparel items using a customized convolutional neural network. As disclosed, f-scores can be generated for apparel items. Training images are identified, where each training image is associated with a corresponding set of tags including information about a plurality of attributes. A first convolutional neural network (CNN) is trained based on the plurality of training images and a first attribute. The first CNN is iteratively refined by, for each respective attribute, removing a set of neurons from the first CNN and retraining the first CNN based on the training imagesand the respective attribute. Upon determining that the first CNN has been trained based on each of the attributes, one or more CNNs are generated based on the first CNN. An image is received, where the image depicts an apparel item. The image is processed using the one or more CNNs, and an f-score for the apparel item is determined based on the output.

Description

Background technique [0001] The present disclosure relates to convolutional neural networks, and more particularly to utilizing a customized convolutional neural network to perform cognitive analysis of images of apparel items. [0002] Many recent advances in commerce (online commerce in particular) are often attributable to back-end analytics enabled by platforms, which help understand market demand and allow managers to fine-tune their offers to better suit customer needs and expectations. One of the fastest growing segments of the trade today encompasses fashion apparel, including accessories, footwear, clothing, jewelry, and more. Clothing item sales are a highly personalized trade category and typically involve very low conversion rates and high return rates. Due to the low conversion rate and highly personalized nature of purchases, suppliers simply cannot tailor their offers to each individual customer. There is a need for systems and methods that cognitively analyze...

AI Technical Summary

Problems solved by technology

Vendors simply cannot tailor their offer to each individual customer due to the low conversion rate and highly personalized nature of purchases

Images