S

Segformer B4 Finetuned Ade 512 512

Developed by nvidia
SegFormer is a Transformer-based semantic segmentation model fine-tuned on the ADE20k dataset, suitable for image segmentation tasks.
Downloads 15.61k
Release Time : 3/2/2022

Model Overview

This model features a hierarchical Transformer encoder and a lightweight all-MLP decoder head, delivering outstanding performance in semantic segmentation tasks.

Model Features

Hierarchical Transformer Architecture
Employs a hierarchical Transformer encoder to effectively capture multi-scale features.
Lightweight Decoder Head
Uses an all-MLP decoder head design to maintain high performance while reducing computational load.
512x512 Resolution Support
The model is fine-tuned at 512x512 resolution, making it suitable for high-resolution image processing.

Model Capabilities

Image Semantic Segmentation
Scene Understanding
Pixel-level Classification

Use Cases

Scene Understanding
House Scene Parsing
Perform semantic segmentation on house images to identify different architectural elements.
Castle Scene Parsing
Perform semantic segmentation on castle images to identify architectural structures and environmental elements.
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase