Swinv2 Tiny Patch4 Window8 256 Finetuned THFOOD 50

🚀 swinv2-tiny-patch4-window8-256-finetuned-THFOOD-50

This model is a fine - tuned version of microsoft/swinv2-tiny-patch4-window8-256 on the THFOOD-50 dataset. It can effectively perform tasks on the THFOOD - 50 dataset, achieving high accuracy in food image classification.

🚀 Quick Start

This model can be used directly through the Hugging Face Transformers library. You can load the model and perform inference tasks.

✨ Features

• Fine - tuned: Based on the pre - trained model [microsoft/swinv2-tiny-patch4-window8-256], it is fine - tuned on the THFOOD-50 dataset, improving the accuracy of food image classification.
• High accuracy: Achieves high accuracy on the training, validation, and test sets.

📦 Installation

Since this model is based on the Transformers library, you can install the necessary libraries using the following command:

pip install transformers datasets torch

💻 Usage Examples

Here is a simple example of using this model for inference:

from transformers import AutoFeatureExtractor, AutoModelForImageClassification
import torch
from PIL import Image
import requests

# Load feature extractor and model
feature_extractor = AutoFeatureExtractor.from_pretrained("your_model_name")
model = AutoModelForImageClassification.from_pretrained("your_model_name")

# Load an image
url = 'https://huggingface.co/datasets/thean/sample_images/resolve/main/FriedChicken.jpg'
image = Image.open(requests.get(url, stream=True).raw)

# Preprocess the image
inputs = feature_extractor(images=image, return_tensors="pt")

# Perform inference
with torch.no_grad():
    logits = model(**inputs).logits

# Get the predicted label
predicted_label = logits.argmax(-1).item()
print(model.config.id2label[predicted_label])

📚 Documentation

Model Performance

This model achieves the following results on different datasets:

• Train set
  • Loss: 0.1669
  • Accuracy: 0.9557
• Validation set
  • Loss: 0.2535
  • Accuracy: 0.9344
• Test set
  • Loss: 0.2669
  • Accuracy: 0.9292

Training and Evaluation Data

The model is trained and evaluated on the THFOOD-50 dataset. However, more detailed information about the data is yet to be provided.

Training Procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e - 05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.3

📄 License

This model is licensed under the afl - 3.0 license.