Best_MobileNetV2_Model

Scrubaroo - MobileNetV2 Waste Density Classifier

A MobileNetV2-based image classifier for waste density detection, trained as part of the Scrubaroo Web3 citizen reward system.

Model Description

• Architecture: MobileNetV2 (pre-trained on ImageNet)
• Task: 3-class image classification (High / Low / NoWaste)
• Training Dataset: KMeans-MobileNetV2 (3,059 images)
• Best Accuracy: 91.34%
• Training Strategy: Weighted CrossEntropyLoss + Data Augmentation

Classes

Usage

import torch
import torchvision.transforms as transforms
import torchvision.models as models
from PIL import Image

# Load model
model = models.mobilenet_v2(weights=None)
model.classifier[1] = torch.nn.Linear(1280, 3)
model.load_state_dict(torch.load('MNV2_WL_TK-MobileNet_V2.pth', 
                                  map_location='cpu'))
model.eval()

# Transform
transform = transforms.Compose([
    transforms.Resize((224, 224)),
    transforms.ToTensor(),
    transforms.Normalize([0.485, 0.456, 0.406],
                         [0.229, 0.224, 0.225])])

# Predict
image = Image.open('image.jpg').convert('RGB')
tensor = transform(image).unsqueeze(0)

with torch.no_grad():
    output = model(tensor)
    predicted = output.argmax(dim=1).item()
    class_names = {0: 'High', 1: 'Low', 2: 'NoWaste'}
    print(class_names[predicted])

Training Details

• Optimizer: Adam (lr=0.001)
• Scheduler: ReduceLROnPlateau (patience=2, factor=0.5)
• Epochs: 10
• Frozen Layers: features[0:16]
• Trainable: features[-3:] + classifier

Related Resources

• Live App: https://scrubaroo.vercel.app/

• GitHub: https://github.com/panoik12/Scrubaroo

• GitHub (ML Code):** https://github.com/panoik12/Scrubaroo_ML_Models

• Paper: Scrubaroo: AI-Powered Waste Detection and Blockchain-Based Citizen Reward System, SETN 2026