ApexVision is a smartphone app that gives people who are blind or have low vision real-time guidance about the world in front of them - identifying obstacles, describing surroundings, and helping them move safely through a space, with results spoken in under 200 milliseconds.
Why it matters: 2.2 billion people worldwide live with vision impairment. Existing tools force a hard tradeoff - dedicated hardware costs $3,500–$4,500, popular apps stream a user's live camera to remote strangers, and cloud-based tools add seconds of delay that make them unsafe at a street crossing. Roughly half of users abandon assistive devices altogether because visible hardware publicly marks them as disabled. ApexVision removes that tradeoff.
How it works: Four computer-vision models - semantic segmentation, depth estimation, object detection, and a vision-language model - run entirely on the user's existing iPhone, with no internet connection and no data ever leaving the device. This makes the experience real-time, fully private, and functional offline. Three modes serve different needs: Finder locates a specific object on request, Safety highlights walkable paths and flags hazards, and Descriptive narrates the surrounding scene on demand.
Proof to date: ApexVision is live on iOS. The underlying technology originated in University of Tennessee computer-vision research, is published in IEEE CVPR, and won the NVIDIA AI City Challenge. It is protected by U.S. Provisional Patent 63/917,447. We completed an institutional pilot with Signal Centers, an assistive-technology center in Chattanooga, which confirmed the real-time navigation capability was new to their program and valued by users. ApexAI was selected for Cohort 1 of the NSF-funded NobleReach Science to Venture program, and completed NSF I-Corps customer discovery.
Next steps: We are expanding to newer LiDAR-equipped devices and to Android, growing institutional pilots with state vocational rehabilitation agencies and assistive-technology centers, and developing a predictive safety layer that anticipates hazards and conveys its own confidence. We are raising a $1M seed round and seeking partners in accessibility distribution, institutional procurement, and edge-AI hardware.
The University of Tennessee, Knoxville is a comprehensive public land‑grant research university—the flagship of the UT System—classified as R1 and serving more than 40,000 students. Industry engagement is anchored by the UT Research Park at Cherokee Farm, where corporate R&D and joint university–national lab facilities sit just across the river from campus, including assets such as the Volkswagen Innovation Hub and an AT&T 5G testbed. UT’s long‑standing partnership with Oak Ridge National Laboratory—via UT‑Battelle and the UT–Oak Ridge Innovation Institute—gives companies streamlined access to national lab capabilities, talent, and joint programs. A statewide Extension network and established co‑op programs connect companies to faculty expertise and student talent across Tennessee and into federal labs. Research is supported by competitive federal sponsors such as the National Science Foundation and the U.S. Department of Energy. Commercialization is managed by the University of Tennessee Research Foundation, which handles IP, licensing, and startup formation.