Models / shi-labs / OneFormer ADE20k DiNAT Large

OneFormer ADE20k DiNAT Large

JSON →
vision

A large-scale transformer-based panoptic segmentation model trained on ADE20k with a Dilated Neighborhood Attention Transformer backbone.

imagevision
Specs
Context & limits
context window
max output
Lifecycle
Dates
released2023-06-01
Resources