license: apache-2.0 tags: - generated_from_trainer metrics: - accuracy model-index: - name: iraqi_foods102 results: [] datasets: - Falah/food102-iraqi-rice-meal language: - en author: Falah G. Salieh location: Iraq, Baghdad

iraqi_foods102

This model is a fine-tuned version of microsoft/swin-base-patch4-window7-224 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5399
• Accuracy: 0.8548

Dataset for Food-102 (Food101+Iraqi-rice-male)

Dataset Name: Food-102

Dataset Summary: Food-102 is an updated version of the Food-101 dataset, now expanded to include 102 food categories. It consists of a total of 102,000 images, with 750 training images and 250 manually reviewed test images provided for each category. The dataset aims to enable food classification tasks and provide a diverse range of food images for research and development purposes. The training images in Food-102 have intentionally not been cleaned, allowing for some level of noise, such as intense colors and occasional mislabeled images. All images in the dataset have been rescaled to have a maximum side length of 512 pixels.

Additional Information:

• Number of Categories: 102
• Total Images: 101,100
• Training Images per Category: 75,825
• Test Images per Category: 25,275
• Image Noise: The training images may contain some noise, including intense colors and occasional mislabeled images.
• Image Rescaling: All images in the dataset have been resized to have a maximum side length of 512 pixels.

Note:

The newly added category "Iraqi rice male food" is not specifically mentioned as part of the Food-101 dataset. If you require further details or have any specific questions about the dataset, please let me know.

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:

• learning_rate: 1e-05
• train_batch_size: 16
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.1273	1.0	592	0.7230	0.8165
0.7414	2.0	1185	0.5696	0.8478
0.5882	3.0	1776	0.5399	0.8548

Framework versions

• Transformers 4.27.1
• Pytorch 2.0.1+cu118
• Datasets 2.9.0
• Tokenizers 0.13.3

Citation

If you use this model in your research, please cite the following paper: