vit-base-xray-pneumonia Open Source Model - Accurately Classify Pneumonia in Chest X-rays with Over 90% Accuracy

Home

Vit Base Xray Pneumonia

Developed by nickmuchi

A ViT-based chest X-ray pneumonia classification model, fine-tuned on pneumonia dataset with 90.06% accuracy

Image Classification

Transformers

Open Source License:Apache-2.0 #Chest X-ray Pneumonia Detection #Medical Image Classification #ViT Fine-tuned Model

Downloads 40

Release Time : 3/9/2022

Model Overview

This model is a fine-tuned version of Google's ViT-base model on a chest X-ray pneumonia dataset, primarily used for classifying pneumonia vs. normal X-ray images.

Model Features

High Accuracy

Achieves 90.06% classification accuracy on the evaluation set

ViT-based Architecture

Uses Vision Transformer architecture, suitable for medical image processing

Few-shot Learning

Adapts to specific medical image classification tasks through fine-tuning pre-trained models

Model Capabilities

Chest X-ray Classification

Pneumonia Detection

Medical Image Analysis

Use Cases

Medical Diagnosis

Pneumonia Assisted Diagnosis

Helps doctors quickly identify pneumonia patients' X-ray images

90.06% accuracy

Medical Image Screening

Used for preliminary screening of large-scale chest X-rays

🚀 vit-base-xray-pneumonia

This model is a fine - tuned version of google/vit-base-patch16-224-in21k on the chest-xray-pneumonia dataset. It can be used for image classification tasks, specifically for detecting pneumonia in chest X - rays.

🚀 Quick Start

This model is a fine - tuned version of google/vit-base-patch16-224-in21k on the chest-xray-pneumonia dataset. It achieves the following results on the evaluation set:

Loss: 0.3387
Accuracy: 0.9006

📚 Documentation

Model Information

Property	Details
Model Type	Fine - tuned version of google/vit-base-patch16-224-in21k
Training Data	chest-xray-pneumonia dataset
Metrics	Accuracy

Training Procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0002
train_batch_size: 16
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e - 08
lr_scheduler_type: linear
num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.1233	0.31	100	1.1662	0.6651
0.0868	0.61	200	0.3387	0.9006
0.1387	0.92	300	0.5297	0.8237
0.1264	1.23	400	0.4566	0.8590
0.0829	1.53	500	0.6832	0.8285
0.0734	1.84	600	0.4886	0.8157
0.0132	2.15	700	1.3639	0.7292
0.0877	2.45	800	0.5258	0.8846
0.0516	2.76	900	0.8772	0.8013
0.0637	3.07	1000	0.4947	0.8558
0.0022	3.37	1100	1.0062	0.8045
0.0555	3.68	1200	0.7822	0.8285
0.0405	3.99	1300	1.9288	0.6779
0.0012	4.29	1400	1.2153	0.7981
0.0034	4.6	1500	1.8931	0.7308
0.0339	4.91	1600	0.9071	0.8590
0.0013	5.21	1700	1.6266	0.7580
0.0373	5.52	1800	1.5252	0.7676
0.001	5.83	1900	1.2748	0.7869
0.0005	6.13	2000	1.2103	0.8061
0.0004	6.44	2100	1.3133	0.7981
0.0004	6.75	2200	1.2200	0.8045
0.0004	7.06	2300	1.2834	0.7933
0.0004	7.36	2400	1.3080	0.7949
0.0003	7.67	2500	1.3814	0.7917
0.0004	7.98	2600	1.2853	0.7965
0.0003	8.28	2700	1.3644	0.7933
0.0003	8.59	2800	1.3137	0.8013
0.0003	8.9	2900	1.3507	0.7997
0.0003	9.2	3000	1.3751	0.7997
0.0003	9.51	3100	1.3884	0.7981
0.0003	9.82	3200	1.3831	0.7997

💻 Usage Examples

Example Images

Pneumonia Chest X - Ray

Pneumonia

Normal Chest X - Ray

Normal

Framework versions

Transformers 4.17.0
Pytorch 1.10.0+cu111
Datasets 1.18.4
Tokenizers 0.11.6

📄 License

This model is licensed under the Apache 2.0 license.

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