My Trash Classification

🚀 Trash Image Classification using Vision Transformer (ViT)

This repository implements an image classification model with a pre - trained Vision Transformer (ViT) from Hugging Face. It classifies images into six categories: cardboard, glass, metal, paper, plastic, and trash.

🚀 Quick Start

This project focuses on trash image classification using a pre - trained Vision Transformer (ViT) model. The model is fine - tuned on a specific dataset to classify images accurately.

✨ Features

• Utilizes a pre - trained Vision Transformer (ViT) model from Hugging Face.
• Fine - tunes the model to classify trash images into six categories.
• Provides a simple inference script for easy usage.

📦 Installation

The provided README does not contain specific installation steps, so this section is skipped.

💻 Usage Examples

Basic Usage

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

url = 'https://cdn.grid.id/crop/0x0:0x0/700x465/photo/grid/original/127308_kaleng-bekas.jpg'
image = Image.open(requests.get(url, stream=True).raw)

model_name = "tribber93/my-trash-classification"
model = AutoModelForImageClassification.from_pretrained(model_name)
processor = AutoImageProcessor.from_pretrained(model_name)
inputs = processor(image, return_tensors="pt")

outputs = model(**inputs)
predictions = torch.argmax(outputs.logits, dim=-1)
print("Predicted class:", model.config.id2label[predictions.item()])

📚 Documentation

Dataset

The dataset consists of images from six categories from garythung/trashnet with the following distribution:

• Cardboard: 806 images
• Glass: 1002 images
• Metal: 820 images
• Paper: 1188 images
• Plastic: 964 images
• Trash: 274 images

Model

We utilize the pre - trained Vision Transformer model google/vit-base-patch16-224-in21k from Hugging Face for image classification. The model is fine - tuned on the dataset to achieve optimal performance. The trained model is accessible on Hugging Face Hub at: tribber93/my-trash-classification

Results

After training, the model achieved the following performance:

🔧 Technical Details

The provided README does not contain specific technical details, so this section is skipped.

📄 License

The provided README does not contain license information, so this section is skipped.