Experimenting with stereo matching
- data folder: contains images and depth results
- My results are not optimal, have a lot of noise and are slow I'll work on
- reducing noise
- optimizing computing time
- My results are not optimal, have a lot of noise and are slow I'll work on
- Stereo folder contains my attmpt at trying to organize the "taxonomy" of stereo in classes for an easy pipeline (It's rushed)
- Costs: different cost functions
- Aggregation: only fixed window for now
- Disparity computing : the different global and energy minimization algorithms besides WTA
- Disparity refinment: nothing yet
- utilities : data loading and visulisation
- I was greatly inspired by https://github.com/2b-t/stereo-matching for their jit implementation but wasn't able to replicate it well