Define your own classes for zero-shot segmentation. Enter class names separated by commas.
๐ Click the examples below to explore!
Examples
Input image
Class names (comma-separated)
Monocular depth and surface normals estimation using a DPT (Dense Prediction Transformer) head on top of a frozen TIPS v2 vision encoder. Trained on the NYU Depth V2 dataset.
๐ Click the examples below to explore!
Examples
Semantic segmentation using a DPT (Dense Prediction Transformer) head on top of a frozen TIPS v2 vision encoder. Trained on ADE20K (150 classes).