Open joyhuang9473 opened 2 years ago
We propose an unsupervised, mid-level representation for a generative model of scenes. our network learns to associate different blobs with different entities in a scene and to arrange these blobs to capture scene layout.
We propose an unsupervised, mid-level representation for a generative model of scenes.
our network learns to associate different blobs with different entities in a scene and to arrange these blobs to capture scene layout.
BlobGAN: Spatially Disentangled Scene Representations