H

Hiera Large 224 Hf

Developed by facebook
Hiera is a hierarchical vision Transformer model that is fast, powerful, and concise, surpassing existing technologies in image and video tasks while being faster.
Downloads 532
Release Time : 5/12/2024

Model Overview

Hiera is a hierarchical vision Transformer model designed for image classification, feature extraction, or masked image modeling. This specific checkpoint is designed for feature extraction.

Model Features

Hierarchical Design
Uses different spatial resolutions and feature dimensions at different network stages through a hierarchical structure to improve efficiency.
Concise Architecture
Removes redundant modules found in traditional vision Transformers, maintaining a concise and efficient architecture.
Efficient Training
Teaches the model to learn spatial bias through MAE training rather than manually adding it via complex architectures.
High Performance
Achieves state-of-the-art performance in multiple image and video recognition tasks while running faster.

Model Capabilities

Image classification
Feature extraction
Masked image modeling

Use Cases

Computer Vision
Image Classification
Used for standard image classification tasks
Performs excellently on benchmarks like ImageNet-1K
Feature Extraction
Extracts multi-level feature representations of images
Can be used for downstream vision tasks
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase