PVSDNet: Joint Depth Prediction and View Synthesis via Shared Latent Spaces in Real-Time

This HuggingFace model currently contains the PVSDNet models of our main work PVSDNet: Joint Depth Prediction and View Synthesis via Shared Latent Spaces in Real-Time. The models here are translation only with encoding set to 2x3xm.

• Please head to the Project Page to see supplementary materials and more details about the project
• Head to our huggingface demo page to see the PVSDNet and PVSDNet-Depth-Only variants in action.