Malong Technologies Announces Winners of the CVPR 2019 iMaterialist Challenge on Product Recognition

LONG BEACH, Calif., June 17, 2019 /PRNewswire/ -- Artificial intelligence leader Malong Technologies today announced the results of the 2019 iMaterialist Challenge on Product Recognition at the Sixth Annual Workshop on Fine-Grained Visual Categorization (FGVC6) at CVPR 2019.

Malong organized the first product recognition challenge ever at CVPR and created the dataset, which is the largest in history in terms of size and scale of product diversity, with more than one million images, more than two thousand stock-keeping units (SKUs) and more than 200 images per SKU.

96 AI teams from around the world participated in the Challenge, submitting a total of 1600 entries from April 1 to the conclusion of the event, June 2, 2019. Winners of the 2019 iMaterialist Challenge are as follows:

    --  1(st) place: JD AI Research
    --  2(nd) place: MTDP-VIC
    --  3(rd) place: EB Lab

"Congratulations to the winners of the iMaterialist Challenge on Product Recognition, and our thanks to everyone who competed," said Matt Scott, co-founder and chief technology officer, Malong Technologies. "Our hope is the challenge will introduce the task of product recognition to the global computer vision research community, helping raise interest, awareness and inspire more research in algorithms to advance progress in our field."

More information on the Challenge is available here:

As computer vision systems become more prevalent in industry, especially in retail, they must recognize products based on images at the SKU level accurately, in real-time and at large-scale. However, the degree of difficulty keeps increasing along with a growing number of SKUs and a shrinking of subtle differences among SKUs - as, for example, different flavors of a soft drink might look the same, or images of a product or SKU look different under different lighting conditions or perspectives. Product photos that a consumer might post online - known as user-generated content, or UGC - can vary greatly cross domain; that is, from photos of the same product taken by a professional (professional-generated content, or PGC).

About Malong Technologies:
Malong Technologies is a global thought leader in AI for product recognition. Since its founding in 2014, the company has focused on advanced deep learning research and development in product recognition for retail applications, with numerous scientific achievements along the way; including inventing the CurriculumNet weakly supervised learning algorithm which won the inaugural WebVision Challenge of CVPR17 by a large margin (relative error rate of nearly 50% between first and second place). Malong Technologies is on a mission to help the retail industry transform with AI to increase efficiency, quality and safety. It maintains strategic alliances with Accenture and Microsoft, among many partnerships with global technology leaders, and it was recently named a Technology Pioneer by the World Economic Forum and a Gartner Cool Vendor for AI in Computer Vision. For more information on Malong Technologies, visit

In China:
Kelly Kang

Kevin Kutz

View original content:

SOURCE Malong Technologies