Model description

Hey everyone! This is the first model I’ve deployed :D. This emotion recognition model is a fine-tuned version of google/vit-base-patch16-224-in21k, trained on the ImageFolder dataset. As a first-timer, I’m proud that this model has achieved such accuracy. I plan to further train it to improve its accuracy. Wish me luck!

Intended uses & limitations

I strongly suggest using an input picture with a clear indication of emotion, as I’ve found that the model can sometimes misinterpret the output. Additionally, this model seems to lack confidence in identifying emotions, as evidenced by the slightly varying scores.

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e-05
train_batch_size: 4
eval_batch_size: 4
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
No log	1.0	160	1.5920	0.375
No log	2.0	320	1.4689	0.4313
No log	3.0	480	1.3699	0.4625
1.4989	4.0	640	1.2204	0.5813
1.4989	5.0	800	1.2019	0.5437
1.4989	6.0	960	1.2126	0.55
0.9362	7.0	1120	1.1846	0.5563
0.9362	8.0	1280	1.2819	0.5312
0.9362	9.0	1440	1.2583	0.525
0.5396	10.0	1600	1.1571	0.6

Framework versions

Transformers 4.41.1
Pytorch 2.3.0+cu121
Datasets 2.19.1
Tokenizers 0.19.1

pratapaadii/image_classification

image_classification