Previous workshops: CVPR 2022, CVPR 2021, CVPR 2020
The rapid development in computer vision and machine learning has caused a
major disruption in the retail industry in recent years. In addition to the rise of
the web and online shopping, traditional markets also quickly embrace AI-related
technology solutions at the physical store level.
Following the introduction of computer vision to the world of retail a new
set challenges emerged in both the physical and online domains. The physical
domain exhibits challenges such as shopper and product interaction detection,
detection of products in crowded store displays, fine-grained classification of
many visually similar classes, as well as dynamically adapting to changes in
data in terms of class appearance variation over time, and new classes that
may appear in the images before they are labeled in the dataset. The online
domain contains similar challenges, but with their own twist. Product search
and recognition is performed on more than 100,000 classes, and also incorporates
textual captions describing the products, and text by users during their search.
All of these challenges are at the heart of the computer vision community, and this workshop aims to present the progress in these challenges and encourage the forming of a community for retail computer vision.
Challenge
AliProducts3: large-scale multi-modal product retrieval
The growing customer demand for E-commerce is becoming more and more
diversified, growing the need for methods that not only require a single modality
such as product images, but also call for the usage of detailed textual captions
that describe said images, as well as more ambiguous texts by users searching for
products. Bridging the gap between visual representation and high-level semantic
concepts remains an open research topic for obtaining users’ search intentions.
The AliProducts3 dataset is designed to bridge this gap. This realistic large-scale
and multi-model dataset will consist of ∼5M samples regarding ∼100K unique
fine-grained products. Participants must find the top-K product candidates to
match a text-and-image query such as ”blue men’s turtleneck sweater”, together
with an image of a specific sweater. This challenge is a natural continuation of
our previous AliProducts2 challege, in which search was done only by text. The
challenge last year had 674 competing teams from all over the world.
CVPR 2023 Challenge Website
Challenge Important Dates
March 22, 2023, 10:00 UTC+8
Registration opens
March 31, 2023, 10:00 UTC+8
Entire training data released
May 31, 2023, 10:00 UTC+8
Registration deadline and test data released
June 19, 2023
CVPR 2023 Workshop
