Cognitive system for IoT based Product authentication in CPG industry
Business Requirement
A US based disruptive start-up needed a real time cognitive system to authenticate a range of food products like edible oils, alcohol etc. Core capabilities we brought in included managing Data across Industrial IoT Devices, Image Processing, Spectrometry, Advanced AI/ML algorithms and Cloud.
Our Solution
Predictive models using anomaly detection paradigm were built to determine authenticity of a range of consumer brand products using AI/ML techniques and were deployed over cloud. The end-to-end cognitive system we built for the client consists of a proprietary Industrial IoT device – a miniaturized Raman Spectrometer, a mobile application to communicate with the spectrometer, and a processing system deployed over AWS cloud. The system uses AI technology at the core to predict product authenticity in near real time. The system collects structured and unstructured spectrometer data through a mobile application to build Predictive Models which are deployed on AWS. Also part of the system was a cloud-based Decision Engine that utilized the AWS deployed predictive models to make predictions and communicate results back to the user on the mobile device.
Business Impact
The new system significantly eliminated human-centric workflow to less than 5% of existing levels, and delivered breakthrough competitive advantage over alternate solutions currently available.
Based on the discovery by Indian Nobel laureate C V Raman, the spectral analysis combined with AI, ML, Deep Learning and Image processing techniques helped to move expensive laboratory testing into the field for instant molecular authentication. This low-cost, advanced, quick and non-invasive use of technology has created a disruptive AI based solution for our client to address a large $1 billion plus market for an effective solution that provides end-to-end visibility across the supply chain for global product brands.
