Self-Supervised Vision Model Improves Underwater Robot Monocular Depth Estimation Accuracy
Researchers developed MAU-Depth, a multi-attention self-supervised monocular depth estimation framework incorporating underwater optical properties and knowledge distillation.

