Problem Statement
The client wanted to evaluate and build an innovative proof of concept to see how machine learning and analytics can help visually impaired passengers to identify the door of a train at a train station
Solution Overview
Aiding Visually Challenged: Explore feasibility of using a combination of Video Analytics using Deep Learning Mobile phone to aid visually challenged passengers to locate the train door Neural networks trained on GPUs on cloud Inference running on iPhone (analytics on edge) Solution built-in data lab
Outcomes
94% accuracy in recognizing open doors of specific train models in real-time Tactile feedback on the mobile app to alert passenger when passing open doors
