Tracking objects across multiple Cameras with Metropolis Microservices

Key Takeaways

managing and automating infrastructure like retail cities warehouses is a challenging problem it often covers thousands of square feet of space and many camera views solving it requires accurate detection and tracking it's anonymized personal identifier information metadata management and more with the reference application for multi-camera tracking from Metropolis microservices we track and re-identify objects anonymously across cameras multi-camera tracking application can be broken down into three key components detection and single camera tracking which happens closer to where the cameras are located then taking the intelligence from Individual sensors fusing them together to track objects across cameras and then finally storage and output of the results now let's see this application in action in this demo we're tracking Shoppers across four cameras in a simulated retail environment this can be used for Shopper analytics or can be used as a foundational blocks for building autonomous stores as we focus on camera number three we can see sharper number five show up on cam 3 and momentarily on cam 4 with the same ID number as he walks in the back aisle we can see him transition from camera number three to camera number two now if I zoom in to the floor plan view I can see the trajectory of the person as he walks through the store this can provide a lot of Shopper insights to businesses now let's look at a journey of another Shopper here we're tracking Shopper number nine as he navigates the aisles of the retail store in addition to retail multi-camera tracking application can be used in smart cities Warehouse Logistics and Industrial use cases this environment is synthetically generated in omnivores which can be used for testing your application or for simulating what-if scenarios which are oftentimes difficult to create in real life