deploy

.gitignore

@@ -0,0 +1,2 @@
+__pycache__
+.venv

.gradio/flagged/dataset1.csv

@@ -1,2 +0,0 @@
-image,output,timestamp
-.gradio/flagged/image/b71cfb0c013a7fad3f6a/header_chiffre_0.jpg,"{""label"": ""Sydney Harbour Bridge"", ""confidences"": [{""label"": ""Sydney Harbour Bridge"", ""confidence"": 0.997890055179596}, {""label"": ""Eiffel Tower"", ""confidence"": 0.9969549179077148}, {""label"": ""Forth Bridge"", ""confidence"": 0.905195415019989}, {""label"": ""Brooklyn Bridge"", ""confidence"": 0.9011015892028809}, {""label"": ""Monumento a la Revolucion"", ""confidence"": 0.8767676949501038}]}",2025-01-23 20:53:44.732426

README.md

@@ -1,12 +1,36 @@
----
-title: Landmark Classification Vit
-emoji: 🌍
-colorFrom: pink
-colorTo: pink
-sdk: gradio
-sdk_version: 5.13.0
-app_file: app.py
-pinned: false
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Landmark Classification with ViT
+
+This repository contains a Gradio-based application for classifying famous landmarks using a ViT (Vision Transformer) model. It includes code for loading landmark categories, managing examples, customizing the Gradio interface, and using a pre-trained model from the Hugging Face Hub.
+
+## Features
+- ViT-based landmark classification
+- Organized categories with ordered data loading
+- Custom theme using Gradio’s theming framework
+- Example images section for quick testing
+
+## Installation
+1. Clone the repository.
+2. Create and activate a virtual environment:
+  ```
+  python -m venv .venv
+  source .venv/bin/activate
+  ```
+3. Install dependencies:
+  ```
+  pip install -r requirements.txt
+  ```
+
+## Usage
+1. Run the main application:
+  ```
+  python app.py
+  ```
+2. Access the Gradio interface in a web browser at the URL printed in the console.
+
+## Project Structure
+- `app.py` : Application entry point
+- `classifier.py` : Model loading and inference logic
+- `custom_theme.py` : Custom theme definitions for Gradio
+- `src/ui/interface.py` : UI layout and Gradio components
+- `src/utils/examples.py` : Example data loading and management
+- `landmark_categories.json` : Configuration of landmark categories

app.py

@@ -1,18 +1,22 @@
-import gradio as gr
+"""
+Main application entry point for the Landmark Classification system.
+Initializes components and launches the Gradio interface.
+"""
+
 from classifier import LandmarkClassifier
 from custom_theme import custom_theme
+from src.utils.examples import load_categories, create_example_data
+from src.ui.interface import create_interface
 
-# Initialize classifier
-classifier = LandmarkClassifier()
-
-# Create interface with Label component configured for dictionaries
-iface = gr.Interface(
-    fn=classifier.classify_image,
-    inputs=gr.Image(type="pil"),
-    outputs=gr.Label(num_top_classes=5), 
-    title="Landmark Image Classification",
-    description="Upload an image to identify famous landmarks.",
-    theme=custom_theme  # Apply the custom theme
-)
+def main():
+    # Initialize components
+    classifier = LandmarkClassifier()
+    categories = load_categories()
+    category_examples = create_example_data(categories)
+    
+    # Create and launch interface
+    iface = create_interface(classifier, category_examples, custom_theme)
+    iface.launch()
 
-iface.launch()
+if __name__ == "__main__":
+    main()

classifier.py

@@ -3,13 +3,21 @@ import torch
 
 class LandmarkClassifier:
     def __init__(self, model_name="bsvaz/landmark-classification-vit"):
+        # Load pre-trained model and processor from HuggingFace hub
         self.model = AutoModelForImageClassification.from_pretrained(model_name)
         self.processor = AutoImageProcessor.from_pretrained(model_name)
 
     def classify_image(self, image):
+        # Preprocess image using the model's required format
         inputs = self.processor(image, return_tensors="pt")
+        
+        # Perform inference without gradient calculation for efficiency
         with torch.no_grad():
             outputs = self.model(**inputs)
+        
+        # Convert logits to probabilities using softmax
         probabilities = torch.nn.functional.softmax(outputs.logits, dim=-1)
         scores = probabilities[0].tolist()
+        
+        # Map class indices to label names and their corresponding probabilities
         return {self.model.config.id2label[i]: score for i, score in enumerate(scores)}

custom_theme.py

@@ -1,18 +1,20 @@
 import gradio as gr
 
-# Define a custom theme with purple as the main color and yellow for contrast
+# Create custom theme with purple and yellow color scheme for visual appeal and accessibility
 custom_theme = gr.themes.Default(
-    primary_hue="purple",
-    secondary_hue="yellow",
-    neutral_hue="gray",
+    # Set main color scheme
+    primary_hue="purple",      # Primary brand color
+    secondary_hue="yellow",    # Accent color for contrast
+    neutral_hue="gray",        # Neutral colors for text and backgrounds
 ).set(
-    body_background_fill='*primary_100',  # Light purple background
-    body_text_color='*neutral_900',       # Dark gray text for contrast
-    button_primary_background_fill='*primary_500',  # Purple buttons
-    button_primary_text_color='white',              # White text on buttons
-    button_primary_background_fill_hover='*primary_600',  # Darker purple on hover
-    slider_color='*primary_500',                    # Purple slider
-    checkbox_background_color='*primary_500',       # Purple checkbox
-    input_background_fill='*neutral_100',           # Light gray input background
-    input_border_color='*primary_500',              # Purple input border
+    # Customize specific UI elements
+    body_background_fill='*primary_100',            # Light purple background for visual comfort
+    body_text_color='*neutral_900',                 # Dark text for readability
+    button_primary_background_fill='*primary_500',  # Distinctive purple buttons
+    button_primary_text_color='white',              # High contrast button text
+    button_primary_background_fill_hover='*primary_600',  # Darker purple for hover state
+    slider_color='*primary_500',                    # Consistent purple for interactive elements
+    checkbox_background_color='*primary_500',       # Match checkbox style with theme
+    input_background_fill='*neutral_100',           # Light background for input areas
+    input_border_color='*primary_500',              # Purple borders for visual hierarchy
 )