inference.py

import torch
import torch.nn as nn
import torch.nn.functional as F
from torchvision import transforms
from PIL import Image

class PlantDiseaseClassifier(nn.Module):
    def __init__(self, num_classes, dropout_rate=0.3):
        super(PlantDiseaseClassifier, self).__init__()
        
        # Use EfficientNet as backbone
        from torchvision import models
        self.backbone = models.efficientnet_b2(pretrained=False)
        
        # Get feature dimension
        num_features = self.backbone.classifier[1].in_features
        
        # Replace classifier with custom head
        self.backbone.classifier = nn.Identity()
        
        # Attention mechanism
        self.attention = nn.Sequential(
            nn.AdaptiveAvgPool2d(1),
            nn.Flatten(),
            nn.Linear(num_features, num_features // 4),
            nn.ReLU(),
            nn.Linear(num_features // 4, num_features),
            nn.Sigmoid()
        )
        
        # Custom classifier head
        self.classifier = nn.Sequential(
            nn.Dropout(dropout_rate),
            nn.Linear(num_features, 512),
            nn.BatchNorm1d(512),
            nn.ReLU(),
            nn.Dropout(dropout_rate * 0.5),
            nn.Linear(512, 256),
            nn.BatchNorm1d(256),
            nn.ReLU(),
            nn.Dropout(dropout_rate * 0.3),
            nn.Linear(256, num_classes)
        )
    
    def forward(self, x):
        features = self.backbone.features(x)
        pooled = F.adaptive_avg_pool2d(features, 1)
        pooled = torch.flatten(pooled, 1)
        attention_weights = self.attention(features)
        attended_features = pooled * attention_weights
        output = self.classifier(attended_features)
        return output

def load_model(model_path):
    checkpoint = torch.load(model_path, map_location='cpu')
    num_classes = len(checkpoint['class_names'])
    model = PlantDiseaseClassifier(num_classes=num_classes)
    model.load_state_dict(checkpoint['model_state_dict'])
    model.eval()
    return model, checkpoint['class_names']

def predict_image(image_path, model, class_names):
    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])
    ])
    
    image = Image.open(image_path).convert('RGB')
    image_tensor = transform(image).unsqueeze(0)
    
    with torch.no_grad():
        outputs = model(image_tensor)
        probabilities = F.softmax(outputs, dim=1)[0]
    
    # Get top predictions
    top_probs, top_indices = torch.topk(probabilities, 3)
    
    results = []
    for i in range(len(top_indices)):
        results.append({
            "label": class_names[top_indices[i].item()],
            "score": top_probs[i].item()
        })
    
    return results