app.py

import os
import numpy as np
import torch
import torch.nn as nn
import gradio as gr
from torchvision.models import efficientnet_v2_m, EfficientNet_V2_M_Weights
from torchvision.ops import nms, box_iou
import torch.nn.functional as F
from torchvision import transforms
from PIL import Image, ImageDraw, ImageFont, ImageFilter
from data_manager import get_dog_description
from urllib.parse import quote
from ultralytics import YOLO
import asyncio
import traceback


model_yolo = YOLO('yolov8l.pt')  


dog_breeds = ["Afghan_Hound", "African_Hunting_Dog", "Airedale", "American_Staffordshire_Terrier", 
              "Appenzeller", "Australian_Terrier", "Bedlington_Terrier", "Bernese_Mountain_Dog", 
              "Blenheim_Spaniel", "Border_Collie", "Border_Terrier", "Boston_Bull", "Bouvier_Des_Flandres", 
              "Brabancon_Griffon", "Brittany_Spaniel", "Cardigan", "Chesapeake_Bay_Retriever", 
              "Chihuahua", "Dandie_Dinmont", "Doberman", "English_Foxhound", "English_Setter", 
              "English_Springer", "EntleBucher", "Eskimo_Dog", "French_Bulldog", "German_Shepherd", 
              "German_Short-Haired_Pointer", "Gordon_Setter", "Great_Dane", "Great_Pyrenees", 
              "Greater_Swiss_Mountain_Dog", "Ibizan_Hound", "Irish_Setter", "Irish_Terrier", 
              "Irish_Water_Spaniel", "Irish_Wolfhound", "Italian_Greyhound", "Japanese_Spaniel", 
              "Kerry_Blue_Terrier", "Labrador_Retriever", "Lakeland_Terrier", "Leonberg", "Lhasa", 
              "Maltese_Dog", "Mexican_Hairless", "Newfoundland", "Norfolk_Terrier", "Norwegian_Elkhound", 
              "Norwich_Terrier", "Old_English_Sheepdog", "Pekinese", "Pembroke", "Pomeranian", 
              "Rhodesian_Ridgeback", "Rottweiler", "Saint_Bernard", "Saluki", "Samoyed", 
              "Scotch_Terrier", "Scottish_Deerhound", "Sealyham_Terrier", "Shetland_Sheepdog", 
              "Shih-Tzu", "Siberian_Husky", "Staffordshire_Bullterrier", "Sussex_Spaniel", 
              "Tibetan_Mastiff", "Tibetan_Terrier", "Walker_Hound", "Weimaraner", 
              "Welsh_Springer_Spaniel", "West_Highland_White_Terrier", "Yorkshire_Terrier", 
              "Affenpinscher", "Basenji", "Basset", "Beagle", "Black-and-Tan_Coonhound", "Bloodhound", 
              "Bluetick", "Borzoi", "Boxer", "Briard", "Bull_Mastiff", "Cairn", "Chow", "Clumber", 
              "Cocker_Spaniel", "Collie", "Curly-Coated_Retriever", "Dhole", "Dingo", 
              "Flat-Coated_Retriever", "Giant_Schnauzer", "Golden_Retriever", "Groenendael", "Keeshond", 
              "Kelpie", "Komondor", "Kuvasz", "Malamute", "Malinois", "Miniature_Pinscher", 
              "Miniature_Poodle", "Miniature_Schnauzer", "Otterhound", "Papillon", "Pug", "Redbone", 
              "Schipperke", "Silky_Terrier", "Soft-Coated_Wheaten_Terrier", "Standard_Poodle", 
              "Standard_Schnauzer", "Toy_Poodle", "Toy_Terrier", "Vizsla", "Whippet", 
              "Wire-Haired_Fox_Terrier"]

class MultiHeadAttention(nn.Module):

    def __init__(self, in_dim, num_heads=8):
        super().__init__()
        self.num_heads = num_heads
        self.head_dim = max(1, in_dim // num_heads)
        self.scaled_dim = self.head_dim * num_heads
        self.fc_in = nn.Linear(in_dim, self.scaled_dim)
        self.query = nn.Linear(self.scaled_dim, self.scaled_dim)
        self.key = nn.Linear(self.scaled_dim, self.scaled_dim)
        self.value = nn.Linear(self.scaled_dim, self.scaled_dim)
        self.fc_out = nn.Linear(self.scaled_dim, in_dim)

    def forward(self, x):
        N = x.shape[0]
        x = self.fc_in(x)
        q = self.query(x).view(N, self.num_heads, self.head_dim)
        k = self.key(x).view(N, self.num_heads, self.head_dim)
        v = self.value(x).view(N, self.num_heads, self.head_dim)

        energy = torch.einsum("nqd,nkd->nqk", [q, k])
        attention = F.softmax(energy / (self.head_dim ** 0.5), dim=2)

        out = torch.einsum("nqk,nvd->nqd", [attention, v])
        out = out.reshape(N, self.scaled_dim)
        out = self.fc_out(out)
        return out

class BaseModel(nn.Module):
    def __init__(self, num_classes, device='cuda' if torch.cuda.is_available() else 'cpu'):
        super().__init__()
        self.device = device
        self.backbone = efficientnet_v2_m(weights=EfficientNet_V2_M_Weights.IMAGENET1K_V1)
        self.feature_dim = self.backbone.classifier[1].in_features
        self.backbone.classifier = nn.Identity()

        self.num_heads = max(1, min(8, self.feature_dim // 64))
        self.attention = MultiHeadAttention(self.feature_dim, num_heads=self.num_heads)

        self.classifier = nn.Sequential(
            nn.LayerNorm(self.feature_dim),
            nn.Dropout(0.3),
            nn.Linear(self.feature_dim, num_classes)
        )

        self.to(device)

    def forward(self, x):
        x = x.to(self.device)
        features = self.backbone(x)
        attended_features = self.attention(features)
        logits = self.classifier(attended_features)
        return logits, attended_features


num_classes = 120
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
model = BaseModel(num_classes=num_classes, device=device)

checkpoint = torch.load('best_model_81_dog.pth', map_location=torch.device('cpu'))
model.load_state_dict(checkpoint['model_state_dict'])

# evaluation mode
model.eval()

# Image preprocessing function
def preprocess_image(image):
    # If the image is numpy.ndarray turn into PIL.Image
    if isinstance(image, np.ndarray):
        image = Image.fromarray(image)

    # Use torchvision.transforms to process images
    transform = transforms.Compose([
        transforms.Resize((224, 224)),
        transforms.ToTensor(),
        transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
    ])

    return transform(image).unsqueeze(0)


def get_akc_breeds_link(breed):
    base_url = "https://www.akc.org/dog-breeds/"
    breed_url = breed.lower().replace('_', '-')
    return f"{base_url}{breed_url}/"


async def predict_single_dog(image):
    image_tensor = preprocess_image(image)
    with torch.no_grad():
        output = model(image_tensor)
        logits = output[0] if isinstance(output, tuple) else output
        probabilities = F.softmax(logits, dim=1)
        topk_probs, topk_indices = torch.topk(probabilities, k=3)
        top1_prob = topk_probs[0][0].item()
        topk_breeds = [dog_breeds[idx.item()] for idx in topk_indices[0]]
        
        # Calculate relative probabilities for display
        raw_probs = [prob.item() for prob in topk_probs[0]]
        sum_probs = sum(raw_probs)
        relative_probs = [f"{(prob/sum_probs * 100):.2f}%" for prob in raw_probs]
        
    return top1_prob, topk_breeds, relative_probs
    

async def detect_multiple_dogs(image, conf_threshold=0.3, iou_threshold=0.45):
    results = model_yolo(image, conf=conf_threshold, iou=iou_threshold)[0]
    dogs = []
    boxes = []
    for box in results.boxes:
        if box.cls == 16:  # COCO dataset class for dog is 16 
            xyxy = box.xyxy[0].tolist()
            confidence = box.conf.item()
            boxes.append((xyxy, confidence))
    
    if not boxes:
        dogs.append((image, 1.0, [0, 0, image.width, image.height]))
    else:
        nms_boxes = non_max_suppression(boxes, iou_threshold)
        
        for box, confidence in nms_boxes:
            x1, y1, x2, y2 = box
            w, h = x2 - x1, y2 - y1
            x1 = max(0, x1 - w * 0.05)
            y1 = max(0, y1 - h * 0.05)
            x2 = min(image.width, x2 + w * 0.05)
            y2 = min(image.height, y2 + h * 0.05)
            cropped_image = image.crop((x1, y1, x2, y2))
            dogs.append((cropped_image, confidence, [x1, y1, x2, y2]))
    
    return dogs


def non_max_suppression(boxes, iou_threshold):
    keep = []
    boxes = sorted(boxes, key=lambda x: x[1], reverse=True)
    while boxes:
        current = boxes.pop(0)
        keep.append(current)
        boxes = [box for box in boxes if calculate_iou(current[0], box[0]) < iou_threshold]
    return keep

    
def calculate_iou(box1, box2):
    x1 = max(box1[0], box2[0])
    y1 = max(box1[1], box2[1])
    x2 = min(box1[2], box2[2])
    y2 = min(box1[3], box2[3])
    
    intersection = max(0, x2 - x1) * max(0, y2 - y1)
    area1 = (box1[2] - box1[0]) * (box1[3] - box1[1])
    area2 = (box2[2] - box2[0]) * (box2[3] - box2[1])
    
    iou = intersection / float(area1 + area2 - intersection)
    return iou


async def process_single_dog(image):
    top1_prob, topk_breeds, relative_probs = await predict_single_dog(image)
    
    # Case 1: Low confidence - unclear image or breed not in dataset
    if top1_prob < 0.15:
        initial_state = {
            "explanation": "The image is unclear or the breed is not in the dataset. Please upload a clearer image of a dog.",
            "image": None,
            "is_multi_dog": False
        }
        return initial_state["explanation"], None, initial_state

    breed = topk_breeds[0]
    
    # Case 2: High confidence - single breed result
    if top1_prob >= 0.45:
        description = get_dog_description(breed)
        formatted_description = format_description(description, breed)
        initial_state = {
            "explanation": formatted_description,
            "image": image,
            "is_multi_dog": False
        }
        return formatted_description, image, initial_state
        
    # Case 3: Medium confidence - show top 3 breeds with relative probabilities
    else:
        breeds_info = ""
        for i, (breed, prob) in enumerate(zip(topk_breeds, relative_probs)):
            description = get_dog_description(breed)
            formatted_description = format_description(description, breed)
            breeds_info += f"\n\nBreed {i+1}: **{breed}** (Confidence: {prob})\n{formatted_description}"

        initial_state = {
            "explanation": breeds_info,
            "image": image,
            "is_multi_dog": False
        }
        return breeds_info, image, initial_state
        

async def predict(image):
    if image is None:
        return "Please upload an image to start.", None, None

    try:
        if isinstance(image, np.ndarray):
            image = Image.fromarray(image)

        dogs = await detect_multiple_dogs(image)
        color_list = ['#FF0000', '#00FF00', '#0000FF', '#FFFF00', '#00FFFF', '#FF00FF', '#800080', '#FFA500']
        annotated_image = image.copy()
        draw = ImageDraw.Draw(annotated_image)
        font = ImageFont.load_default()

        dogs_info = ""

        for i, (cropped_image, detection_confidence, box) in enumerate(dogs):
            color = color_list[i % len(color_list)]
            draw.rectangle(box, outline=color, width=3)
            draw.text((box[0] + 5, box[1] + 5), f"Dog {i+1}", fill=color, font=font)
        
            top1_prob, topk_breeds, relative_probs = await predict_single_dog(cropped_image)
            combined_confidence = detection_confidence * top1_prob
            
            dogs_info += f'<div class="dog-info" style="border-left: 5px solid {color}; margin-bottom: 20px; padding: 15px;">'
            dogs_info += f'<h2>Dog {i+1}</h2>'
            
            if combined_confidence < 0.2:
                dogs_info += "<p>The image is unclear or the breed is not in the dataset. Please upload a clearer image.</p>"
                
            elif top1_prob >= 0.45:
                breed = topk_breeds[0]
                description = get_dog_description(breed)
                dogs_info += format_description_html(description, breed)
                
            else:
                dogs_info += "<h3>Top 3 possible breeds:</h3>"
                for breed, prob in zip(topk_breeds, relative_probs):
                    description = get_dog_description(breed)
                    dogs_info += f"<div class='breed-section'>"
                    dogs_info += f"<h4>{breed} (Confidence: {prob})</h4>"
                    dogs_info += format_description_html(description, breed)
                    dogs_info += "</div>"
            
            dogs_info += '</div>'

        html_output = f"""
        <style>
        .dog-info {{ 
            border: 1px solid #ddd; 
            margin-bottom: 20px; 
            padding: 15px; 
            border-radius: 5px; 
            box-shadow: 0 2px 5px rgba(0,0,0,0.1); 
        }}
        .dog-info h2 {{ 
            background-color: #f0f0f0; 
            padding: 10px; 
            margin: -15px -15px 15px -15px; 
            border-radius: 5px 5px 0 0; 
        }}
        .breed-section {{
            margin-bottom: 20px;
            padding: 10px;
            background-color: #f8f8f8;
            border-radius: 5px;
        }}
        </style>
        {dogs_info}
        """

        initial_state = {
            "dogs_info": dogs_info,
            "image": annotated_image,
            "is_multi_dog": len(dogs) > 1,
            "html_output": html_output
        }
        
        return html_output, annotated_image, initial_state

    except Exception as e:
        error_msg = f"An error occurred: {str(e)}\n\nTraceback:\n{traceback.format_exc()}"
        print(error_msg)
        return error_msg, None, None


def show_details_html(choice, previous_output, initial_state):
    if not choice:
        return previous_output, gr.update(visible=True), initial_state

    try:
        breed = choice.split("More about ")[-1]
        description = get_dog_description(breed)
        formatted_description = format_description_html(description, breed)
        
        html_output = f"""
        <div class="dog-info">
            <h2>{breed}</h2>
            {formatted_description}
        </div>
        """
        
        initial_state["current_description"] = html_output
        initial_state["original_buttons"] = initial_state.get("buttons", [])
        
        return html_output, gr.update(visible=True), initial_state
    except Exception as e:
        error_msg = f"An error occurred while showing details: {e}"
        print(error_msg)
        return f"<p style='color: red;'>{error_msg}</p>", gr.update(visible=True), initial_state


def format_description_html(description, breed):
    html = "<ul style='list-style-type: none; padding-left: 0;'>"
    if isinstance(description, dict):
        for key, value in description.items():
            html += f"<li style='margin-bottom: 10px;'><strong>{key}:</strong> {value}</li>"
    elif isinstance(description, str):
        html += f"<li>{description}</li>"
    else:
        html += f"<li>No description available for {breed}</li>"
    html += "</ul>"
    akc_link = get_akc_breeds_link(breed)
    html += f'<p><a href="{akc_link}" target="_blank">Learn more about {breed} on the AKC website</a></p>'
    return html


with gr.Blocks() as iface:
    gr.HTML("<h1 style='text-align: center;'>🐶 Dog Breed Classifier 🔍</h1>")
    gr.HTML("<p style='text-align: center;'>Upload a picture of a dog, and the model will predict its breed and provide detailed information!</p>")
    gr.HTML("<p style='text-align: center; color: #666; font-size: 0.9em;'>Note: This is an AI model and predictions may not always be 100% accurate. The model provides its best estimates based on training data.</p>")
    
    
    with gr.Row():
        input_image = gr.Image(label="Upload a dog image", type="pil")
        output_image = gr.Image(label="Annotated Image")
    
    output = gr.HTML(label="Prediction Results")
    initial_state = gr.State()
    
    input_image.change(
        predict,
        inputs=input_image,
        outputs=[output, output_image, initial_state]
    )
    
    gr.Examples(
        examples=['Border_Collie.jpg', 'Golden_Retriever.jpeg', 'Saint_Bernard.jpeg', 'French_Bulldog.jpeg', 'Samoyed.jpg'],
        inputs=input_image
    )
    
    gr.HTML('For more details on this project and other work, feel free to visit my GitHub <a href="https://github.com/Eric-Chung-0511/Learning-Record/tree/main/Data%20Science%20Projects/Dog_Breed_Classifier">Dog Breed Classifier</a>')

if __name__ == "__main__":
    iface.launch()