M

Mask2former Swin Tiny Ade Semantic

Developed by facebook
Mask2Former is a unified image segmentation model based on Transformer, capable of handling instance segmentation, semantic segmentation, and panoptic segmentation tasks.
Downloads 7,834
Release Time : 1/5/2023

Model Overview

Mask2Former employs a unified paradigm to solve various image segmentation tasks by predicting a set of masks and their corresponding labels. Compared to its predecessor MaskFormer, it offers improvements in both performance and efficiency.

Model Features

Unified Segmentation Paradigm
Unifies instance segmentation, semantic segmentation, and panoptic segmentation into an instance segmentation approach
Efficient Attention Mechanism
Replaces traditional pixel decoders with multi-scale deformable attention Transformer
Masked Attention Decoder
Uses Transformer decoder with masked attention to improve performance without increasing computational cost
Efficient Training Method
Enhances training efficiency by calculating loss based on sampled points rather than full masks

Model Capabilities

Semantic Segmentation
Instance Segmentation
Panoptic Segmentation
Image Analysis

Use Cases

Computer Vision
Scene Understanding
Identify and segment objects in complex scenes
Accurately identifies and segments 150 semantic categories in the ADE20k dataset
Autonomous Driving
Object detection and segmentation in road scenes
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase