Visual Product Search and Tagging APIs Designed for Industry Verticals
Menswear Jacket Minimalist
Light Green Short-Sleeves College Style
Leisure Easy-Going
Handsome Denim Euro Style


Visual Product Search and Tagging APIs Designed for Industry Verticals
Ceiling Light


ProductAI® provides state-of-the-art APIs for visual product recognition based on artificial intelligence.

The ProductAI® cloud platform equips businesses with image retrieval and auto-tagging for industry verticals such as fashion, furniture, cars, textiles and wine. Based on cutting-edge deep learning and computer vision R&D, developed by a world-class team of scientists and engineers hailing from Microsoft Research, Google and Oxford's famed VGG lab, product recognition with market-leading accuracy is now accessible to any developer for pennies on the dollar.


以图搜图@1x Created with Sketch.

Vertical AI Image Retrieval

This technology enables developers to get a private API to their own high-performance product image search engine. A leap ahead of generic "reverse image search" solutions which look at all images in the same way, ProductAI® utilizes deep image retrieval models which are architected and trained specifically for select industries. This means that images are indexed with domain-specific semantics and are thus of significantly higher performance than a generic solution. For example, the ProductAI® fashion vertical custom search can enable street to shop scenarios with an index and detector aware of apparel style and materials, robust to lighting and angle changes. Build your own world-class product search app in minutes.

图像识别@1x Created with Sketch.

Product Image Auto-Tagging

This technology enables developers to automatically annotate product images with relevant information just like a human labeler can do, but with sub-second speed and a fraction of the cost. By just inputting an image which contains a product, ProductAI® will generate in real-time, highly relevant textual keywords utilizing specialized deep learning models adapted to select product verticals. Additionally, ProductAI® auto-tagging also offers object detection, e.g. will provide not just textual annotation, but also the spatial location information of the tag within the image. Finally, the auto-tagging doesn't just provide a single concept label per image, but multi-concept labels as well. Build your own world-class product tagger app in minutes.

"AI+" Scenarios

ProductAI® enables product recognition for the following AI+ industry vertical scenarios. Mouse over to learn more.



We believe ProductAI® offers the highest accuracy and speed of any product recognition PaaS for the vertical industries it serves. We back this statement up from both academic and industrial investigations as of April 2017. From the academic side, we have tested our technology against the relevant public benchmarks and found no published research paper with higher accuracies reported. From the industry side, we have tested our technology against competing services and have yet to find a player that outperforms what we offer. In terms of speed, we have tested our image retrieval with various verticals up to index sizes of 20M and have yet to record more than half of 1 second of processing time on our servers.

How we can achieve this? Unique Tech. Exclusive Data. World-Class Talents. In terms of technology, we have invented new deep learning algorithms that enable significant performance gains. In terms of data, we have trained our models on 75+M unique data training samples. In terms of talents, our team consists of elite scientists and engineers from some of the top research institutions in the world.




ProductAI® has been battle-tested in Asia by over 1 billion API requests in production by Chinese and Japanese users during the first 9 months of going online. As of today, the platform has grown to become the market-leading AI platform for select verticals in the region. Taking one example, the textile industry: 7 out of the top-10 biggest fabric companies in China trust ProductAI® and have integrated the API into key areas of their businesses, such as e-commerce for their end-users, and their backend systems for their enterprise operations. Other market leading companies, such as the #1 fashion app in China, the #1 stock photography provider in China, and other major retailers and product makers trust and build with ProductAI® for their AI needs.

ProductAI® is designed with security and privacy as top priorities from the ground up. Note that the entirety of ProductAI® for the US Market, everything from data, models, and services, are all 100% running on highly secure servers in the United States. Privacy and security are of the utmost importance and part of our core values as a company - not a single datum of customer data is ever transmitted or made accessible to anyone other than the customer themselves.


ProductAI® API is elegantly designed for software engineers by software engineers. Besides being able to call the simple yet secure Restful API directly over HTTPS, language-specific SDKs have been created for Python, PHP, Java and C#, and open sourced on GitHub. We believe in the beauty of clean, well-documented and intuitive interfaces. We care deeply about the productivity of the developers that are building with ProductAI® and welcome pull requests and issues for our SDKs from the community.

Part of the appeal of ProductAI® is that the price of highly accurate AI services can also be highly inexpensive. Feel free to experiment and build large-scale services such as an image search engine up to millions of images without fear of breaking the bank. And as we continue to invent new algorithms that reduce our own infrastructure costs, we pass on those savings to our customers.



We are an award-winning artificial intelligence startup whose mission is to empower developers and businesses with the most accurate visual product recognition APIs available on the market. In 2014, the company, known as “Malong” in China, was founded by Matthew Scott and Dr. Dinglong Huang. Prior to co-founding, Dr. Dinglong Huang, CEO, was a VP at TripAdvisor, with previous experience at Microsoft and Google. Matthew Scott, CTO, was a senior technical staff member of Microsoft Research, with 15+ years R&D experience in computer vision and machine learning.

In 2015, the company graduated with honors from the Microsoft Ventures Startup Accelerator, won the first “AI Pioneer” award from Microsoft in 2016 and received other top tier awards for its AI technology from Amazon and NVIDIA. Malong has a world-class team of research scientists, including ImageNet / CVPR workshop competition winners (e.g. "Olympic medalists” of the computer vision field). The company has raised over $10 million in VC funding.


ProductAI® is currently available for the US market in private beta. Apply now for a free trial invitation.