June 15, 2021: PhD Thesis Pre-submission Seminar by Bikash Santra, Senior Research Fellow, ECSU

On automatic identification of retail products in images of racks in the supermarkets


Bikash Santra

Senior Research Fellow, ECSU

Date: Tuesday, June 15, 2021
Time: 3pm
Venue: Online (through google meet, https://meet.google.com/zwm-mfwi-gdj)



An image of a rack in a supermarket displays a number of retail products. The identification and localization of these individual products from the images of racks is a challenge for any machine vision system. In this thesis, we suggest a set of solutions for automatic identification of region proposals representing these retail products. First, we develop an end-to-end machine vision system for detection and localization. The proposed system introduces a novel exemplar-driven region proposal strategy that overcomes the shortcomings of traditional exemplar-independent region proposal schemes like selective window search. Second, we design a novel classifier that essentially differentiates the similar yet non-identical (fine-grained) products for improving the performance of our machine vision system. The proposed fine-grained classifier simultaneously captures both part-level and object-level cues of the products for accurately distinguishing the fine-grained products. Third, we propose a graph-based non- maximal suppression strategy that addresses an important bottleneck of conventional greedy non-maximal suppression algorithm for disambiguation of overlapping region proposals generated in an intermediate step of our proposed system. Finally, we find the empty regions (or gaps between products) in each shelf of any rack by creating a graph of superpixels for the rack. We extract the visual features of superpixels from our graph convolutional and Siamese networks. Subsequently, we send the graph along with the features of superpixels to a structural support vector machine for discovering the empty regions of the shelves. The efficacy of the proposed approaches are established through various experiments on our In-house dataset and three publicly available benchmark datasets: Grozi-120, Grocery Products, and WebMarket.

All are cordially invited to attend.


Dipti Prasad Mukherjee (Supervisor)
Electronics and Communication Sciences Unit