PersonGONE: Image Inpainting for Automated Checkout Solution

Vojtěch Bartl, Jakub Špaňhel, Adam Herout; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2022, pp. 3115-3123

Abstract

In this paper, we present a solution for automatic checkout in a retail store as a part of AI City Challenge 2022. We propose a novel approach that uses the "removal" of unwanted objects -- in this case, body parts of operating staff, which are localized and further removed from video by an image inpainting method. Afterwards, a neural network detector can detect products with a decreased detection false positive rate. A part of our solution is also automatic detection of ROI (the place where products are shown to the system). We reached 0.4167 F1-Score with 0.3704 precision and 0.4762 recall which placed us at the 7th place of AI City Challenge 2022 in corresponding Track 4. The code is made public and available on GitHub.

Related Material

[pdf]
[bibtex]
@InProceedings{Bartl_2022_CVPR, author = {Bartl, Vojt\v{e}ch and \v{S}pa\v{n}hel, Jakub and Herout, Adam}, title = {PersonGONE: Image Inpainting for Automated Checkout Solution}, booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops}, month = {June}, year = {2022}, pages = {3115-3123} }