Image Classification with Pre-trained ResNet50

This project demonstrates how to use a pre-trained ResNet50 model for image classification. It includes downloading the model, loading image data, performing predictions, and saving the results with predicted labels.

Requirements

• Python 3.6+
• PyTorch
• torchvision
• PIL (Pillow)
• requests

Setup

1. Clone the repository:

   git clone https://github.com/your-repo/image-classification-resnet50.git
   cd image-classification-resnet50

2. Install the required packages:

   pip install torch torchvision pillow requests

3. Ensure you have a folder named test_set in the root directory containing the images you want to classify. The folder structure should follow the format expected by ImageFolder, i.e., images organized in subfolders per class (though the class labels won't be used in this example).

Usage

Downloading the Pre-trained Model

The script checks if the ResNet50 model pre-trained on ImageNet is present. If not, it downloads the model.

Transforming the Data

Images are resized to 224x224 pixels, converted to tensors, and normalized using the mean and standard deviation of the ImageNet dataset.

Loading the Dataset

The test_set folder is loaded using torchvision.datasets.ImageFolder, and a DataLoader is created for batch processing.

Evaluating the Model

The compute function evaluates the model on the test images, adds the predicted labels to the images, and saves the results in the output folder.