Storex, The New Age Retail Analytics

Retail runs on analytics. It is critical to know customers for successful retail execution. However very limited data captured about the true potential of stores.

StoreX is the most advanced retail analytics system based on artificial intelligence infrastructure. Using Entegria’s state of the art image processing framework VisionX’s, it detects all demographics around the store. Not only demographics, it is also possible to analyze mood, attention and accessories.

 

Lightining Fast

VisionX with 149 neural network layers and 208 million parameters, it predicts age, age group, gender, mood, attention and accessorries in one call within milliseconds.

 

Visual Identification System

VisionX is an artifical intelligence based image recognition system developed by Entegria. It can detect from cancer cell to any object. In StoreX, we use the customized version of VisionX image recognition framework in order to detect only people, demographics, attention, mood and accessorries.

In retail, as mentioned above knowing customers are critical for success. Many retailers suffer due to lack of visitor demographics of their stores. If known properly, it will be possible to distribute right products with appropriate quantities to stores. As you may guess, true potential of each store is very different and this is not only changing by place, it also change by time. So retailers should have known the number of visitors, their demographics, mood and their attention to their products for each store.

Customized Sales Action Plan For Each Store

StoreX provides all these information in quantity and nature for every single store in any time frames. Retailers can customize their sales action plan per store and plan store allocations based on demographics and interest, organize store regions based on people attention, forecast sales based on statistics and take actions to increase customer satisfaction based on mood data.

 

Unique Visits by Face Recognition

StoreX calculates special face vectors for each detected individual in the images. With this information it is possible to distinguish store staff from visitors.  This information can also be used for analyzing unique visits and visit frequencies.

 

Precise Counting by Visual Tracking System

StoreX uses the most advanced tracking system based on unsupervised learning based clustering. System gives a unique identifier for each individual and keeps this number for every frame till the individual is out of sight.  This unique identifier removes the risk of duplicate counting.  Especially in crowded scenes it works perfectly and keeps the trajectory of each individual precisely

 

Cost Efficiency with One Camera for Multiple Detections

StoreX is not a single task detector. It can detect outside traffic, inside traffic, demographics, attention and moods simultaneously with a single processing unit using one camera. This is very important for small to midsize stores in setting up a cost efficient retail analytics system. If it is possible to capture all events with one camera, then it is possible to analyze all with one StoreX processing unit.

 

More Details with Multiple Camera Support

StoreX can also work with multiple cameras. It can process the images from different sources simultaneously with a very limited fps drop. This architecture will provide retailers more details about their sales area. Depending on the graphics cards processing capabilities, this can be increased to more than 10 different image sources.

 

Centralized Image Processing

StoreX also supports centralized processing. If images from stores can be transferred to a central computing center, it is possible to process all simultaneously. This requires large GPU stacks. StoreX is optimized for NVIDIA graphics cards.

 

Detection Accurracy

 

Traffic in front of Store	Accurracy
Traffic Direction	%95
Count	%95
Gender	%90
Traffic Capture Rate	%90

 

Store Demographics	Accurracy
People Counting	%95
Gender	%90
Age Group	%85
Age	%80
Accessories	%85
Mood	%80
Attention Time	%95