Clear and Look - Using Diffusion Models for depth prediction in bad wheather

As a placeholder, we name the model we will train as:

• MATHPLACEHOLDER0 : Joint dehazing (RGB) and Depth-Prediction (2D Depth Map)
• MATHPLACEHOLDER1 : Only dehaze (RGB)
• MATHPLACEHOLDER2 The "clear" GT-RGB image
• MATHPLACEHOLDER3 The "hazy" RGB image used as input to our models

We will use UniDepthV2 as a basline to compare:

• UniDepthV2($I$) <-> UniDepthV2($\text{CAL}_{\text{dehaze}}(I)$)

These samples can be visualized here.