vit-base-patch16-224-in21k-euroSat Open-Source Model - A Practical Tool for High-Precision Remote Sensing Image Classification

Vit Base Patch16 224 In21k Eurosat

Developed by philschmid

A high-precision remote sensing image classification model fine-tuned on the EuroSAT dataset based on Google's Vision Transformer architecture

Image Classification

Transformers

Open Source License:Apache-2.0 #High-precision image classification #Remote sensing image recognition #EuroSAT dataset

Downloads 28

Release Time : 3/2/2022

Model Overview

This model is an image classification model based on the Vision Transformer (ViT) architecture, specifically fine-tuned for the EuroSAT remote sensing image dataset, achieving a classification accuracy of up to 99.06%.

Model Features

High-precision classification

Achieves 99.06% accuracy and 100% top-3 accuracy on the EuroSAT test set

Based on ViT architecture

Utilizes the Vision Transformer architecture with powerful image feature extraction capabilities

Efficient training

Requires only 5 training epochs to achieve near-perfect classification performance

Model Capabilities

Remote sensing image classification

Multi-category image recognition

High-precision scene classification

Use Cases

Remote sensing analysis

Land use classification

Classify and identify different land types in satellite images

Can accurately identify 10 different land types

Environmental monitoring

Monitor changes in environmental elements such as forests, farmlands, and water bodies

Geographic information systems

Automated map annotation

Automatically identify and annotate geographic features in satellite images

🚀 philschmid/vit-base-patch16-224-in21k-euroSat

This model is a fine - tuned version of google/vit-base-patch16-224-in21k, aiming to provide high - accuracy image classification.

🚀 Quick Start

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

Train Loss: 0.0218
Train Accuracy: 0.9990
Train Top - 3 - accuracy: 1.0000
Validation Loss: 0.0440
Validation Accuracy: 0.9906
Validation Top - 3 - accuracy: 1.0
Epoch: 5

📚 Documentation

Model Information

Property	Details
Model Name	philschmid/vit-base-patch16-224-in21k-euroSat
Base Model	google/vit-base-patch16-224-in21k
Task	Image Classification
Dataset	eurosat
Metrics	accuracy, top - 3 - accuracy
Accuracy	0.9906
Top - 3 - accuracy	1.0000

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

optimizer: {'inner_optimizer': {'class_name': 'AdamWeightDecay', 'config': {'name': 'AdamWeightDecay', 'learning_rate': {'class_name': 'PolynomialDecay', 'config': {'initial_learning_rate': 3e - 05, 'decay_steps': 3585, 'end_learning_rate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}}, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e - 08, 'amsgrad': False, 'weight_decay_rate': 0.01}}, 'dynamic': True, 'initial_scale': 32768.0, 'dynamic_growth_steps': 2000}
training_precision: mixed_float16

Training results

Train Loss	Train Accuracy	Train Top - 3 - accuracy	Validation Loss	Validation Accuracy	Validation Top - 3 - accuracy	Epoch
0.4692	0.9471	0.9878	0.1455	0.9861	0.9998	1
0.0998	0.9888	0.9996	0.0821	0.9864	0.9995	2
0.0517	0.9939	0.9999	0.0617	0.9871	1.0	3
0.0309	0.9971	0.9999	0.0524	0.9878	0.9998	4
0.0218	0.9990	1.0000	0.0440	0.9906	1.0	5

Framework versions

Transformers 4.15.0
TensorFlow 2.7.0
Datasets 1.17.0
Tokenizers 0.10.3

📄 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