Tata Consultancy Services Limited

Shilpa Yadukumar Rao

Pranoy Hari

Dipti Prasad Mukherjee

Avishek Kumar Shaw

Jeisobers Thirunavukkarasu

Rajashree Ramakrishnan

Bikash Santra

Ehtesham Hassan

Anurag Saran

Preeti Johri

Archan Ray

Nishant Kumar

Deepa Jayaveer

Dhanasekaran Gopal

The fine-grained variations in product images are usually due to slight variations in text, size, and color of the package. Both marginal variations in image content and illumination poses an important challenge in product classification. This disclosure relates to a system and method for fine-grained classification of similar-looking products utilizing object-level and part-level information. The system simultaneously captures an object-level and part-level information of the product. The object-level classification score of the product is estimated with the trained RC-Net, a deep supervised convolutional autoencoder. For annotation-free modelling of part-level information of the product the discriminative part-proposal of the product is identified around the BRISK key points. An ordered sequence of the discriminative part-proposals and the product image, encoded using stacked convolutional LSTM network, estimates the part-level classification score. Finally, the trained RC-Net and stacked conv-LSTM network jointly classifies the product image based on the final classification score.

Fine-grained classification of retail products

Most current systems decompose sales volume of a product into base volume and additional volumes associated to each sales driver. Here base volume refers to sales quantity that is derived from non-promoted sales period. It leads base volume of a product with partial price effect. In addition, base volume includes volume contributed by store location, space allocated to product category and assortment etc. In the present disclosure a system and method for sales decomposition is described by computing reference base volume and sales contribution by each sales driver. Here reference base volume is the volume derived by excluding the effect of sales drivers namely, retailer price, competitor price, demographics, weather conditions, space occupied by the category to which the product belongs, number of available substitute products in assortment. Retail strategies are recommended in real time based on sales contribution of the sales driver for the product.

System and method for sales volume decomposition

In online apparel shopping, user finds difficulty in choosing apparel of his/her preferences. This disclosure relates to recommend an apparel specific to a user by estimated attribute values in digital environment. An information associated with the user and apparels is processed to obtain a first parameter and a second parameter and associated data independent matrix and data dependent matrix is derived. The data independent matrix and the data dependent matrix is learned to obtain trained multivariate multi structure model. A significant pattern associated with each user is determined by extracting plurality of images. The derived significant pattern is deployed on k-nearest neighbor model to obtain recommended pattern. The trained multivariate multi structure model and the recommended pattern is applied on state of the user in a real time to estimate the attribute values. The estimated attribute values is mapped with existing apparels in repository to recommend prioritized apparels.

Method, system and medium for user specific apparel attribute recommendation

This disclosure relates generally to a system and method to identify various products on a plurality of images of various shelves of a retail store to facilitate compliance with respect to planograms. Planogram is a visual plan, which designates the placement of products on shelves and merchandising display fixtures of a retail store. Planograms are used to create consistency between store locations, to provide proper shelf space allocation, to improve visual merchandising appeal, and to create product-pairing suggestions. There are a few assumptions considering one instance per product class is available beforehand and the physical dimension of each product template is available in some suitable unit of length. In case of absence of physical dimension of the products, a context information of the retail store will be used. The context information is that the products of similar shapes or classes are arranged together in the shelves for consumers' convenience.

Method and system for region proposal based object recognition for estimating planogram compliance

A method and system is provided for planogram compliance check based on visual analysis. The present method and system comprises acquiring a product shelf image, identifying a region of interest (ROI) from the at least one product shelf image, extracting a perspective distortion rectification of the identified ROI to generate a rectified shelf image, applying Hausdroff distance based image map for occupancy estimation in shelf images extracting one or more rows from the rectified shelf image wherein the rectified shelf image comprises a plurality of horizontal row partitions, and deriving a product count and placement information for each of the extracted one or more rows based on the extracted rows and the occupancy estimation.

Method and system for planogram compliance check based on visual analysis

Object recognition based estimation of planogram compliance provides an expected arrangement of products in shelves. Identifying whether a product is placed in an appropriate location of a shelf is a challenging task due to various real-time parameters associated with image capturing. In the present disclosure, an input image associated with shelf of a retail store is received and a product images are cropped. Further, a set of reference images stored in a database are scaled corresponding to the input image. Further, one or more composite matching scores are calculated based on normalized cross-correlation and shape based feature matching to obtain one or more probable product images associated with a location. Further, a Directed Acyclic Graph (DAG) is constructed based on the one or more composite scores and the one or more probable products. Finally, an optimal matching product image for a particular location is obtained from the DAG.

Growing community of inventors

Chennai, India

Shilpa Yadukumar Rao