Blood Cell Cancer Classifier (ResNet18)

Fine-tuned ResNet18 for classifying blood cell images into 4 classes.

Classes

0. Benign
1. Malignant Pre-B
2. Malignant Pro-B
3. Malignant early Pre-B

Performance

• Validation Accuracy: 99.38%

Usage

from huggingface_hub import PyTorchModelHubMixin
import torch.nn as nn
from torchvision import models

class BloodCancerResNet18(nn.Module, PyTorchModelHubMixin):
    def __init__(self, num_classes=4):
        super().__init__()
        base = models.resnet18(weights=None)
        base.fc = nn.Linear(base.fc.in_features, num_classes)
        self.model = base

    def forward(self, x):
        return self.model(x)

model = BloodCancerResNet18.from_pretrained("your-username/blood-cancer-resnet18")
model.eval()

Training Data

Trained on the Blood Cell Cancer ALL-4class dataset.