Zippin has developed the next generation of checkout-free technology enabling retailers to quickly deploy frictionless shopping in their stores. Their patent-pending approach uses AI, machine learning and sensor fusion technology to create the best consumer experience. Zippin’s vision is to banish checkout lines and self-scanners for good, so shoppers can zip in and out with their purchases.
Problem
The customer wanted to create a ‘checkout-free shopping’ experience for its customers. Zippin built Machine Learning (ML) algorithms which can determine what item(s) were picked by customer A, customer B, customer C, …. In order to determine the final price and process check out.
For this requirement, a lot of training had to be done to their ML algorithms which included Item Annotation, Body Part Annotation (elbows, fingers), Torso Annotation, Activity Labelling, and Exception Handling. These training data-sets can range from tens of thousands to few millions.
The Intellekt Solution
Intellekt put together a team of 25 Computer Vision (CV) annotators to do the various annotations using our Annotation and Workflow Platform tool for Computer Vision. All of the annotators are trained and certified internally on AIML basics and tool. Once the ML model achieved considerable confidence, it was moved to Production. Intellekt then was in an engagement to provide Live Support for Exception Handling. This occurs when the model is unable to identify or act per its input.
Result From The Customer
Over 5.7 million images were annotated till date. There was a 35% Productivity increase after they came to Intellekt. This was achieved because of the platform tools, best practices and processes put in place. We do not just have the annotators putting boxes or annotations, but we understand what is required to support the Data Science teams. We have been working together for 4 years now!
Services Provided
Bounding Box Annotation
Image Annotation
Data Labelling
Model Training
Annotation / Day
%