🏛️ ArchiNet: Architectural Style Classifier

ArchiNet is a deep learning model that classifies images of buildings into different architectural styles — from Baroque and Bauhaus to Gothic and Greek Revival.

Features

• Classifies images into 25 architectural styles.
• Provides top-3 predictions with confidence scores.
• Supports both CPU and GPU inference.
• Optimized ONNX model for fast inference.

Installation & Usage

Run Locally

1. Clone this repository:

   git clone https://github.com/aalizelau/ArchiNet.git
   cd ArchiNet
   

2. Install dependencies:

   pip install -r requirements.txt

3. Run the Gradio app:

   python app.py

Model Details

• Input: 384x384 RGB image (ImageNet normalized)
• Output: Top-k probabilities across 25 styles
• Format: ONNX

Dataset

The model was trained on the Architecture Dataset from Kaggle:
🔗 https://www.kaggle.com/datasets/wwymak/architecture-dataset