The dataset contains images related to bread cancer. Each image has a corresponding mask image.
We selected the following pre-trained models:
- VGG16: Known for its simplicity and effectiveness in image classification tasks.
- ResNet50: Utilizes residual learning, which allows training of very deep networks.
- InceptionV3: Combines multiple filter sizes and network-in-network architectures, providing high accuracy.
We fine-tuned the models by modifying the top layers and training them on the dataset. The following layers were modified:
- GlobalAveragePooling2D: To reduce the feature maps.
- Dense (1024 units): Added for high-level features.
- Dense (1 unit): Output layer for binary classification with sigmoid activation.
To use this project:
- Clone the repository to your local machine.
- Ensure that the dataset is stored at the correct path, as specified in the notebook.
- Open the
Transfer_Learning_Assignment.ipynbnotebook in Jupyter or another IPython notebook viewer. - Run the cells sequentially to perform the data preprocessing, model training, and evaluation.
The models are evaluated based on the following metrics:
- Accuracy
- Loss
- Precision
- Recall
- F1 Score
These metrics are crucial for assessing the performance of the models, particularly in a medical context where accuracy and reliability are paramount.
The results section will contain tables and charts generated from the notebook, providing a visual and quantitative comparison of the performance of the different models.
Contributions to this project are welcome. You can contribute by improving the model training pipeline, introducing new models, or enhancing the data preprocessing steps.
This project is licensed under the MIT License - see the LICENSE file for details.