Leo Qin

← All work

06 · Hackathon

AI White Cane

Group Second Place at the Moonshot 48 Hour Hackathon Cup.

Role
Team Lead
When
Apr 2025 · Shenzhen, China
YOLOv5 camera Amap API location LLM voice pyttsx3 user obstacle about 25 m (estimate)
Camera, location and a language model combine into spoken directions.

Built with

  • Unihiker
  • YOLOv5
  • Amap API
  • LLM
  • pyttsx3

About the project

I led a team at the Moonshot 48 Hour Hackathon Cup, where we took Group Second Place. A Unihiker board spots obstacles within about 25 m (an estimate) using YOLOv5, adds location from the Amap API, and an LLM turns both into directions that pyttsx3 speaks offline, for visually impaired users.

Worth knowing. Maps and the language model need internet, and the board is too weak for real use yet.

Next project