Contains 3 deep neural networks in reverse-rendering from photo graphs. Data sets: Li et al's, Boss et al's, mine.

1. Download Li's network and training it with Li's data
2. Download Boss's network and training with Li's data
3. implement Deschaintre et al's network and training with Li's data
4. test all three network and compare the results
5. Using Boss' data to do Step 1-4
6. Using my data to do Step 1-4
7. Summarise the results

Li: http://cseweb.ucsd.edu/~viscomp/projects/SIGA18ShapeSVBRDF/

Boss: https://github.com/NVlabs/two-shot-brdf-shape

Deschaintre: https://github.com/valentin-deschaintre/Single-Image-SVBRDF-Capture-rendering-loss