How Student Engineers Turned AI Vision into Smart Aid for the Blind

Project Overview

A group of students from Wuhan Engineering University, motivated by a love of technology and social responsibility, set out to design a complete smart‑assistive ecosystem for the visually impaired. In 2023, China’s blind population exceeded 17 million, with many lacking affordable, modern assistive devices.

Recognizing the high cost and limited functionality of traditional canes, the team envisioned intelligent devices that could act as the "eyes" of blind users, converting visual information into understandable feedback.

Technical Implementation

The team developed a smart cane and guide glasses equipped with navigation, obstacle‑avoidance, emergency alert, remote security, and positioning functions. They employed computer‑vision algorithms to capture the environment and used semantic‑segmentation‑based infrared enhancement and image‑fusion techniques to improve detection in low‑light or high‑interference settings.

For deep‑learning training, they adopted Huawei’s MindSpore framework and the ModelArts AI development platform, deploying models efficiently on the cloud. Local training was performed with the Huawei Ascend Atlas200 developer kit, achieving high‑speed, high‑efficiency training.

By leveraging Huawei Cloud resources for dataset annotation, training, and validation, the team reduced training time by over 30% and used the cloud as the backend for their mobile app, significantly lowering communication latency.

Results and Impact

The AI‑enhanced devices provide real‑time voice feedback about the surrounding environment, support path planning and navigation, and improve walking speed and obstacle‑avoidance success rates by more than 18% compared with conventional aids. In emergencies, users can press an alarm button, prompting family members to receive SMS alerts and real‑time location data via the app.

Through Huawei Cloud’s mature development suite, the team streamlined development, reduced code complexity, and achieved faster algorithm iteration, ultimately winning the 2024 Huawei Developer Competition China‑region Student Track first prize.