Starting point: An open source tool with limited capabilities
The customer has an autonomous store solution that enables cashierless checkout. This is based on extensive research based on user interaction in store. They were using an open-source video labeling solution - CVAT - to label things like people, products and people interacting with products. They had started to look for a new platform given the limitations of the existing one: slow, buggy, long turnaround for data processing.
Simplifying Unstructured Data Processing with super.AI
The customer reached out to super.AI given its powerful Unstructured Data Processing (UDP) solution with a well defined API and a simple interface. They were also happy we were able to process multiple data types, not just video. We plugged in the customer’s video footage to our API and were able to process hundreds of hours of video footage for them in an easier and faster way.
We were able to process hundreds of hour of video footage for the customer at a much faster turnaround time than the previous open source tool they were using and better accuracy. In fact, their video processing time decreased 10-15 times from the starting point.
We were generating lot of overhead trying to plug in the proprietary camera data to our open source tool. The tool had a very limited UX and we were getting the data processed quite slowly. Thanks to super.AI’s API import option and its very straightforward user interface, we have found a data processing platform that fits our needs much better.