Open-source image classification model swin-tiny-patch4-window7-224-finetuned-eurosat - Achieve accurate image classification for free

Swin Tiny Patch4 Window7 224 Finetuned Eurosat

Developed by dgrinwald

This is a vision model based on the Swin Transformer Tiny architecture, fine-tuned on the EuroSAT dataset, primarily used for image classification tasks.

Image Classification

Transformers

Open Source License:Apache-2.0 #Remote Sensing Image Classification #Swin Transformer Fine-tuning #High-precision Classification

Downloads 16

Release Time : 7/9/2022

Model Overview

This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 for image classification tasks, achieving an accuracy of 84.65% on the evaluation set.

Model Features

High Accuracy

Achieved a classification accuracy of 84.65% on the evaluation set

Based on Swin Transformer

Utilizes the advanced Swin Transformer architecture with efficient window attention mechanisms

Fine-tuned Model

Optimized for specific tasks based on the foundational model

Model Capabilities

Image Classification

Remote Sensing Image Analysis

Use Cases

Remote Sensing Image Analysis

Land Use Classification

Classify land types in satellite images

Accuracy reached 84.65%

Environmental Monitoring

Identify and classify different environmental zones

🚀 swin-tiny-patch4-window7-224-finetuned-eurosat

This model is a fine - tuned version of microsoft/swin-tiny-patch4-window7-224 on the imagefolder dataset. It can achieve good results in image classification tasks, with an accuracy of 0.8465 on the evaluation set.

📚 Documentation

Model Information

Property	Details
Model Type	Fine - tuned version of microsoft/swin-tiny-patch4-window7-224
Training Data	imagefolder
Metrics	Accuracy: 0.8464730290456431

This model achieves the following results on the evaluation set:

Loss: 0.3266
Accuracy: 0.8465

Training and Evaluation

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e - 05
train_batch_size: 32
eval_batch_size: 32
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 128
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.2941	1.0	17	1.1717	0.4689
1.0655	2.0	34	0.9397	0.5560
0.8008	3.0	51	0.6153	0.7303
0.7204	4.0	68	0.5665	0.7427
0.6931	5.0	85	0.4670	0.7801
0.6277	6.0	102	0.4328	0.8465
0.5689	7.0	119	0.4078	0.8174
0.6103	8.0	136	0.4060	0.8091
0.5501	9.0	153	0.4842	0.7884
0.6018	10.0	170	0.3780	0.8423
0.5668	11.0	187	0.3551	0.8631
0.5192	12.0	204	0.4514	0.8216
0.5133	13.0	221	0.3598	0.8174
0.5753	14.0	238	0.4172	0.8091
0.4833	15.0	255	0.4685	0.8050
0.5546	16.0	272	0.4474	0.7842
0.5179	17.0	289	0.4570	0.7884
0.5017	18.0	306	0.4218	0.8050
0.4808	19.0	323	0.4094	0.8050
0.4708	20.0	340	0.4693	0.7759
0.5033	21.0	357	0.3141	0.8672
0.4859	22.0	374	0.3687	0.8257
0.516	23.0	391	0.3819	0.8216
0.4822	24.0	408	0.3391	0.8506
0.4748	25.0	425	0.3281	0.8506
0.4914	26.0	442	0.3308	0.8631
0.4354	27.0	459	0.3859	0.8133
0.4297	28.0	476	0.3761	0.8133
0.4747	29.0	493	0.2914	0.8672
0.4395	30.0	510	0.3025	0.8548
0.4279	31.0	527	0.3314	0.8506
0.4327	32.0	544	0.4626	0.7842
0.446	33.0	561	0.3499	0.8382
0.4011	34.0	578	0.3408	0.8465
0.4418	35.0	595	0.3159	0.8589
0.484	36.0	612	0.3130	0.8548
0.4119	37.0	629	0.2899	0.8589
0.4453	38.0	646	0.3200	0.8465
0.4074	39.0	663	0.3493	0.8465
0.3937	40.0	680	0.3003	0.8672
0.4222	41.0	697	0.3547	0.8299
0.3922	42.0	714	0.3206	0.8589
0.3973	43.0	731	0.4074	0.8133
0.4118	44.0	748	0.3147	0.8589
0.4088	45.0	765	0.3393	0.8506
0.3635	46.0	782	0.3584	0.8257
0.403	47.0	799	0.3240	0.8506
0.3943	48.0	816	0.3536	0.8216
0.4085	49.0	833	0.3270	0.8465
0.3865	50.0	850	0.3266	0.8465

Framework versions

Transformers 4.20.1
Pytorch 1.11.0+cu113
Datasets 2.3.2
Tokenizers 0.12.1

📄 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