Exploiting Egocentric Vision on Shopping Cart for Out-Of-Stock Detection in Retail Environments

Dario Allegra, Mattia Litrico, Maria Ausilia Napoli Spatafora, Filippo Stanco, Giovanni Maria Farinella

International Workshop on Assistive Computer Vision and Robotics, 2021

In conjuction with ICCV 2021

Giovanni Maria Farinella

Dataset and annotations
Continuous detection and efficient monitoring of Out-of-Stock (OOS) of products in retail environments is a key factor to improve stores profits. Traditional methods require labour-intensive human work dedicated to check for products to refill raising the requirement of automatic solution to detect OOS. We propose a deep learning approach for detection of OOS areas training a Convolutional Neural Network (CNN) to predict attention maps useful to obtain the probability to find OOS in certain areas and hence suggest the retail employers where to intervene. To study the problem we consider the EgoCart dataset extending it with weak labels related to OOS products. To prove the validity of the proposed approach, we evaluated results with both objective measures and a subjective analysis provided by human reviewers on the obtained attention maps. Achieved performance demonstrates that the proposed pipeline is promising to help for the refilling process in retail domain.
The above videoclip shows the performance of the proposed approach. (Top-left) RGB frames of the employed dataset; (Bottom-left) the ground truth attention maps; (Bottom-right) the predicted attention maps; (Top-right) the overlap between frames and predicted maps. Note that the "hot" areas of the heat maps are related to the OOS. We recommend to decrease the playing speed for a better visual checking.