app.py

import gradio as gr

from PIL import Image, ImageOps
import numpy as np
from torchvision import transforms
import torch
import torch.nn as nn
import torch.nn.functional as F
class LargeNet(nn.Module):

    def __init__(self):
        super(LargeNet, self).__init__()
        self.name = "large"
        self.conv1 = nn.Conv2d(3, 5, 5)
        self.pool = nn.MaxPool2d(2, 2)
        self.conv2 = nn.Conv2d(5, 10, 5)
        self.fc1 = nn.Linear(10 * 29 * 29, 32)
        self.fc2 = nn.Linear(32, 8)

    def forward(self, x):
        x = self.pool(F.relu(self.conv1(x)))
        x = self.pool(F.relu(self.conv2(x)))
        x = x.view(-1, 10 * 29 * 29)
        x = F.relu(self.fc1(x))
        x = self.fc2(x)
        x = x.squeeze(1) # Flatten to [batch_size]
        return x
    
def preprocess_image(image, target_size=(128, 128)):
    # Load the image
    # image = Image.open(image_path).convert("RGB")
    image = image.convert("RGB")
    print('image' , image)
    # Maintain aspect ratio and pad
    image = ImageOps.fit(image, target_size, method=Image.BICUBIC, centering=(0.5, 0.5))
    
    # Normalize pixel values (0 to 1) or standardize
    image_array = np.array(image) / 255.0  # Normalize to [0, 1]
    
    return image_array

model = LargeNet()
model.load_state_dict(torch.load("./model_large_bs64_lr0.001_epoch29"))
model.eval()
print(model)
def classify_image(image_path):
    classes = ["Gasoline_Can", "Hammer", "Pebbels", "pliers",
                      "Rope", "Screw_Driver", "Toolbox", "Wrench"]
    image = preprocess_image(image_path)
    image_tensor = torch.tensor(image).permute(2, 0, 1).unsqueeze(0).float()  # Add batch dimension
    print('image ', image_tensor.shape)
    with torch.no_grad():
        outputs = model(image_tensor)
        _, predicted_class = torch.max(outputs, 1)
    print(classes[predicted_class.item()])
    return classes[predicted_class.item()]


transform = transforms.Compose([
    transforms.Resize((128, 128)),
    transforms.RandomHorizontalFlip(),
    transforms.RandomRotation(15),
    transforms.ColorJitter(brightness=0.2, contrast=0.2, saturation=0.2),
    transforms.ToTensor(),  # Convert to PyTorch Tensor
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])  # Standardize
])



# classify_image('rope1.jpeg')
# Gradio interface
demo = gr.Interface(
    fn=classify_image,  # Classification function
    inputs=gr.Image(type="pil"),
    outputs=gr.Textbox(),
    title="Mechanical Tools Classifier"
)

if __name__ == "__main__":
    demo.launch()  # Launch Gradio app