End-to-end trainable mask generation
...D_P is for the pixel-level Domain Adaptation discriminator
...D_F is for the feature-level (latent-vector actually) Domain Adaptation discriminator
Branches
| branch | comment | active |
|---|---|---|
| megadepth | add input depth information using MegaDepth model | Yes (Méli) |
| watchdogs | train on Watchdogs dataset | Yes (Méli) |
| discriminative | segmentation (discriminative) rather than generative approach | Yes |