The primary objective of this project is to develop a suite of AI models specifically designed for the
non-uniform low-lighting and low-visibility conditions found in underground mining. This suite will
feature specialized multi-modal AI models for three core computer vision tasks: object detection, semantic (including instance) segmentation, and depth estimation. Each model will leverage
multi-modal inputs, where images of the scene are captured using different camera sensors such
as RGB and Near Infrared (NIR), and fused within the model space to ensure accurate predictions.
Another objective is to ensure that the AI-driven camera system is highly versatile and capable of
serving multiple use cases such as body-mounted cameras, surveillance cameras, cabin cameras,
and cameras integrated into autonomous (and remote) vehicles.
Accessibility Controls