Publication detail

PersonGONE: Image Inpainting for Automated Checkout Solution

BARTL, V. ŠPAŇHEL, J. HEROUT, A.

Original Title

PersonGONE: Image Inpainting for Automated Checkout Solution

Type

conference paper

Language

English

Original 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.

Keywords

automatic checkout, product counting, image inpainting, object detection, object tracking

Authors

BARTL, V.; ŠPAŇHEL, J.; HEROUT, A.

Released

24. 6. 2022

Publisher

IEEE Computer Society

Location

New Orleans, LA

ISBN

2160-7516

Year of study

2022

Number

7

Pages from

3114

Pages to

3122

Pages count

9

URL

https://ieeexplore.ieee.org/document/9857198

BibTex

@inproceedings{BUT178943,
  author="Vojtěch {Bartl} and Jakub {Špaňhel} and Adam {Herout}",
  title="PersonGONE: Image Inpainting for Automated Checkout Solution",
  booktitle="2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)",
  year="2022",
  series="IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops",
  volume="2022",
  number="7",
  pages="3114--3122",
  publisher="IEEE Computer Society",
  address="New Orleans, LA",
  doi="10.1109/CVPRW56347.2022.00351",
  issn="2160-7516",
  url="https://ieeexplore.ieee.org/document/9857198"
}