Cutting-edge research and development make our solutions the most accurate, insightful, and powerful on the market.
Rebooting retail for the digital age
Harness computer vision, deep learning, and artificial intelligence to sell more in stores.
A view of every SKU
We capture images from a variety of sources and then analyze photos down to the pixel level.Learn More
Our object-recognition and placement-detection engines extract data from images with industry-leading accuracy.Learn More
Business-intelligence algorithms combine image data with sales and strategy data to provide deep insights on KPIs that matter most to you.Learn More
Artificial intelligence automatically prioritizes tasks and identifies the ideal staff to complete each job.Learn More
A view of every SKUBecause every client has different data-collection capabilities, and because each,
retail environment is unique, Trax has developed a suite of image-capture solutions for
capturing every product in every store.
See shelves from all the right angles
Trax gathers images via our mobile app, shelf cameras, dome cameras, and the digital eyes of roving robots. Our hybrid approach allows us to thrive in any retail environment, from drugstores to grocery stores to convenience stores.
Deep-learning technology analyzes images in pixel-precise detail, recognizing the tiniest of differences between SKUs. The powerful recognition engine can overcome poor lighting, background clutter, and obstructions.
Ensuring data qualityTrax is dedicated to the highest data-accuracy standards. Our rigorous quality-assurance system provides transparent information about our performance so that you can make strategic decisions with total confidence. Learn how our Data Quality Enforcement Framework ensures the credibility of your data.
Identifying products and determining their exact placement in a dynamic retail environment is a complex computing task. We’ve spent years perfecting our system so clients can be sure they’re receiving accurate data.
Get high-quality data you can trust
As soon as images are uploaded to our cloud servers, every pixel is analyzed, and products are identified against an extensive database of SKUs. The system checks the placement and arrangement of items, comparing execution to planograms and KPIs.
The quality of our image-recognition data is assured by a robust system of AI and human validations. Thanks to machine learning, our accuracy continues to improve every second of every day.
Man and machineOur computer-vision technology is continuously improved with help from human eyes. Photos are sent to domain experts for “voting”, a process that validates product identifications and trains our active-learning algorithms.
Trax is much more than just an image-recognition platform. We transform shelf data into metrics for monitoring performance, intelligence for beating competitors, and inspiring strategy insights.
Turn data into insights
The Trax platform uses shelf data to measure execution, store compliance, market share, staff and SKU performance, and much, much more. Dashboards can be customized around clients’ KPIs and combined with other data sources, such as EPOS.
Detailed reports reveal how categories, brands and SKUs are trending over time and in different regions, allowing you to make informed big-picture decisions.
For your informationFilter out the noise and focus on the data that matter. Our computer-vision technology measures every aspect of shelf health. Then, reporting tools provide actionable insights based on KPIs, from product availability to planogram compliance.
After gathering, analyzing, and reporting data, we use the insights and our artificial-intelligence tech to improve operations and increase sales in brick-and-mortar stores.
Optimize your store
By combining shelf data with a client’s KPIs, we prioritize the execution tasks that will lead to the greatest successes.
AI also powers the matching algorithm for our on-demand workforce. That way, we always send the very best people for completing clients’ merchandising tasks.
Finally, our location-aware engagement app motivates consumers to engage with brands and stores and rewards their loyalty.
Our CVPR HonorsTrax is proud to be constantly recognized and honored at Conference on Computer Vision and Pattern Recognition (CVPR), an annual conference widely regarded as one of the most important conferences in its field.
RetailVision at CVPR 2021
AI for real-world problems
Detection in densely packed scenes
products recognized every month
total product recognitions to date
Commonly asked questionsOur team is here to help. One of our technology experts would love to answer your questions.
Every day, we take a sample of the images we collect and do a deep analysis of the recognition results. We aggregate the results and provide Accuracy reports every 3 weeks. You can learn more about the process in this video.
The algorithm starts a new project with 90% accuracy and improves to 96% over the first few months, thanks to human validation. Humans continue to quality check to control for packaging updates.
Open architecture is a core tenet of our approach. We have successfully integrated our technology with SFA tools, BI systems, and data lakes. To ensure our solutions work seamlessly with clients’ ecosystems, we work closely with their tech teams on integrations.
A dedicated Project Manager and Analyst make setup, launch, and stabilization painless. To create your image database, we use your existing library or even build it from scratch.
Data ownership is a negotiable term in our contracts. Retailers and brands can own or resell the data we collect or adopt a shared ownership model with Trax.
Trax has successfully passed security due diligence for leading global retailers and is GDPR compliant. Our camera system includes sensor alerts in the event of an occlusion (an obstruction in front of the camera) that contains a human presence. The system analyzes every picture to determine whether a human is present before uploading to Trax’s cloud. If a picture contains a person, it’s immediately discarded. The process is repeated until the camera captures a picture without an occlusion. The occluded images are deleted from the gateway server and the IoT camera device. At the end of the process, no photos of people are saved at the store or the cloud.