vit_small_patch8_224.lunit_dino Open-source Image Classification Model - Free Deployment to Assist in Pathological Image Classification

Vit Small Patch8 224.lunit Dino

Developed by 1aurent

An image classification model based on the Vision Transformer (ViT), trained on 33 million histological sections using the DINO self-supervised learning method, suitable for pathological image classification tasks.

Image Classification Open Source License:Other #Pathological image feature extraction #Self-supervised learning #Medical image analysis

Downloads 167

Release Time : 11/26/2023

Model Overview

This is a Vision Transformer model specifically designed for pathological image classification, which can serve as a basic model for feature extraction and is suitable for the field of medical image analysis.

Model Features

Training on large-scale pathological data

Trained on 33 million histological sections from multiple pathological datasets

Self-supervised learning

Trained using the DINO self-supervised learning method without the need for labeled data

Feature extraction backbone

Can serve as a basic model for feature extraction and is suitable for downstream tasks

Model Capabilities

Pathological image feature extraction

Histological section classification

Medical image analysis

Use Cases

Medical diagnosis

Cancer tissue classification

Identify and classify different types of cancer tissue sections

Pathological image analysis

Extract pathological image features for auxiliary diagnosis

🚀 Model card for vit_small_patch8_224.lunit_dino

A Vision Transformer (ViT) image classification model trained on 33M histology patches from various pathology datasets, offering powerful feature extraction capabilities for image - classification tasks.

✨ Features

Feature Backbone: Serves as a robust feature extraction backbone.
DINO SSL Method: Utilizes the DINO self - supervised learning method.
Model Stats: With 21.7M parameters and designed for 224 × 224 x 3 images.
Multiple Datasets: Trained on diverse pathology datasets like BACH, CRC, MHIST, PatchCamelyon, and CoNSeP.

📦 Installation

No specific installation steps are provided in the original README.

💻 Usage Examples

Basic Usage

from urllib.request import urlopen
from PIL import Image
import timm

# get example histology image
img = Image.open(
  urlopen(
    "https://github.com/owkin/HistoSSLscaling/raw/main/assets/example.tif"
  )
)

# load model from the hub
model = timm.create_model(
  model_name="hf-hub:1aurent/vit_small_patch8_224.lunit_dino",
  pretrained=True,
).eval()

# get model specific transforms (normalization, resize)
data_config = timm.data.resolve_model_data_config(model)
transforms = timm.data.create_transform(**data_config, is_training=False)

output = model(transforms(img).unsqueeze(0))  # output is (batch_size, num_features) shaped tensor

📚 Documentation

Model Details

Property	Details
Model Type	Feature backbone
SSL Method	DINO
Params (M)	21.7
Image sizes	224 × 224 x 3
Papers	Benchmarking Self - Supervised Learning on Diverse Pathology Datasets: https://arxiv.org/abs/2212.04690
Datasets	BACH, CRC, MHIST, PatchCamelyon, CoNSeP
Original	https://github.com/lunit-io/benchmark-ssl-pathology
License	lunit - non - commercial

📄 License

This model is released under the lunit - non - commercial license.

📚 Citation

@inproceedings{kang2022benchmarking,
  author    = {Kang, Mingu and Song, Heon and Park, Seonwook and Yoo, Donggeun and Pereira, Sérgio},
  title     = {Benchmarking Self - Supervised Learning on Diverse Pathology Datasets},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month     = {June},
  year      = {2023},
  pages     = {3344-3354}
}

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご