Higher Eye is an embedded system that uses AI to supervise climbing gyms and active entertainment centers. The system is designed to maximize safety and minimize the risk of accidents. It is monitioring the climbers and raises an alarm if it detects a dangerous situation.
In the most common case, it prevents climbers from climbing without a harness. The system can also detect if there is more than one climber on the same route, or if a climber is reaching out of his lane, as shown on the photo below.
copyright capital.bg
At its essence, the system combines a camera system and a computer vision algorithm. The camera system is installed in the climbing gym and constantly monitors the climbers. The computer vision algorithm analyzes the video, detects a scenario, and makes the system respond to it by triggering an alarm and saving a video stream from the event.
In 2021 I had the idea to integrate Posenet into the E-Walls system, so climbers can select climbing routes using hand gestures, instead of their mobile devices.
I tested a prototype with a hardware from Google Coral and it managed multiple poses quite well, so the idea for Higher Eye was born.
We positioned cameras at the gym's highest points, inspiring the name ‘High Eye.’ To avoid any unwanted associations with ‘drugged eyes’, we added ‘Higher’.
The project development happened during the chip shortage after the COVID-19 pandemic, and the hardware was hard to find. I had to find a new solution and switched to a regular camera and a computer vision algorithm on a dedicated server.
Posenet was not enough for the task, so I had to train a custom model to recognize the climber, the climber`s pose, and the belay rope in one go. Walltopia had access to numerous climbing gyms, so the system became a low-hanging fruit for a new product.
Leveraging a reputation for high-quality products and manufacturing expertise, Techtopia quickly began selling the system to climbing gyms and active entertainment centers.
The system is still in use and is constantly being supported.