intraoralphoto-classifier 0.1.1

Intraoral photo classification with ResNet18

intraoralphoto-classifier

This repository provides a deep learning system for classifying intraoral dental photographs into five standard orthodontic viewpoints:

• center
• up (upper occlusal)
• down (lower occlusal)
• left
• right

The project evaluates preprocessing, data augmentation strategies, multiple image resolutions, and two versions of ResNet-18 (pretrained and trained from scratch).

1. Installation

Install from PyPi:

pip install intraoralphoto-classifier

Install from source:

git clone https://github.com/AImageLab-zip/intraoralphoto-classifier
cd intraoralphoto-classifier
pip install -e .

The framework automatically uses GPU if available; otherwise CPU is used.

2. Inference

from intraoralphoto_classifier.config import get_device
from intraoralphoto_classifier.infer import run_inference

run_inference(
    patient_dir="<path_to_patient_folder>",
    model_path="<best_model.pth>",
    output_dir="<results_folder>",
    device=get_device(),
)

3. Model Weights or Checkpoints

Download the model Here: https://drive.google.com/drive/u/0/folders/1LM9fIciXma-2ak9nyRqOzRrinzIH7K0S