vit-base-patch16-224-in21k-weather-images-classification

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

Loss: 0.2255
Accuracy: 0.9340
Weighted f1: 0.9341
Micro f1: 0.9340
Macro f1: 0.9372
Weighted recall: 0.9340
Micro recall: 0.9340
Macro recall: 0.9354
Weighted precision: 0.9347
Micro precision: 0.9340
Macro precision: 0.9398

Model description

This is a classification model of weather images.

For more information on how it was created, check out the following link: https://github.com/DunnBC22/Vision_Audio_and_Multimodal_Projects/blob/main/Computer%20Vision/Image%20Classification/Multiclass%20Classification/Weather%20Images/Weather_Images_ViT.ipynb

Intended uses & limitations

This model is intended to demonstrate my ability to solve a complex problem using technology.

Training and evaluation data

Dataset Source: https://www.kaggle.com/datasets/jehanbhathena/weather-dataset

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: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Weighted f1	Micro f1	Macro f1	Weighted recall	Micro recall	Macro recall	Weighted precision	Micro precision	Macro precision
2.4333	1.0	337	0.3374	0.9036	0.9028	0.9036	0.9080	0.9036	0.9036	0.9002	0.9088	0.9036	0.9234
0.4422	2.0	674	0.2504	0.9228	0.9226	0.9228	0.9285	0.9228	0.9228	0.9273	0.9248	0.9228	0.9318
0.1051	3.0	1011	0.2255	0.9340	0.9341	0.9340	0.9372	0.9340	0.9340	0.9354	0.9347	0.9340	0.9398

Framework versions

Transformers 4.25.1
Pytorch 1.12.1
Datasets 2.8.0
Tokenizers 0.12.1