ResNetModelFT for Skin Cancer Classification

Model Details

Model Architecture: ResNet-18
Framework: PyTorch
Input Shape: 224x224 RGB images
Number of Parameters: ~11.7M (ResNet-18 pretrained model)
Output: Multi-class classification (9 classes)

Model Description

This model uses ResNet-18, a well-known deep residual network, pre-trained on ImageNet. The model is fine-tuned by replacing the fully connected layer to accommodate multi-class classification for skin cancer detection. Only the fully connected layer is trainable, while the convolutional layers of the ResNet model are frozen to retain pretrained features.

The final model performs multi-class classification with 9 output classes corresponding to different skin cancer types.

Training Details

Optimizer: Adam
Batch Size: 64
Loss Function: Cross-Entropy Loss
Number of Epochs: 10
Dataset: Skin Cancer 9-Class Dataset

Metrics (Validation Set)

Class	Precision	Recall	F1-Score
0	1.00	0.06	0.12
1	0.45	0.31	0.37
2	0.57	0.25	0.35
3	0.00	0.00	0.00
4	0.32	1.00	0.48
5	0.31	0.25	0.28
6	0.50	0.67	0.57
7	0.20	0.06	0.10
8	0.14	1.00	0.24

Overall Accuracy: 0.31
Macro Average Precision: 0.39
Macro Average Recall: 0.40
Macro Average F1-Score: 0.28
Weighted Average Precision: 0.40
Weighted Average Recall: 0.31
Weighted Average F1-Score: 0.25

License

This model is released under the MIT License.

This model has been pushed to the Hub using the PytorchModelHubMixin integration: