Manmo & Karim
Team consisting of an MLOps engineer (Ramify) and Head of IA (POC Innovation), skilled in MLOps, multimodal R&D (vision+audition), cloud deployment (GCloud/NVIDIA).
YouTube Video
Project Description
Our project aimed to design an assistive tool to help visually impaired individuals navigate their surroundings safely and independently. The concept was to use smart glasses equipped with cameras to capture the environment and a haptic feedback system to translate visual information into tactile sensations, allowing users to “feel” their surroundings instead of relying on sound.
During development, we decided to prioritize building the core tool — including environment detection, object recognition, and feedback mechanisms — before implementing the speech-to-text (STT) module. This choice was mainly due to the complexity of integrating real-time video processing within the SDK, which required significant effort and debugging time.
Once the video integration was functional, we planned to revisit the STT system to enable voice-based interactions and controls, improving accessibility and user experience.