| | --- |
| | license: mit |
| | --- |
| | |
| | # SkinSense - Hugging Face Model |
| |
|
| |
|
| | SkinSense is a machine learning model for diagnosing skin diseases based on skin lesion images. It is built using the PyTorch framework and utilizes a fine-tuned ResNet101 architecture. The model can predict whether a skin lesion is benign or malignant, as well as provide a specific diagnosis for malignant lesions. |
| |
|
| | ## Model Overview |
| |
|
| | The SkinSense model is designed to assist medical professionals in diagnosing skin diseases by analyzing images of skin lesions. It was trained on a large dataset of skin lesion images with corresponding labels for diagnosis. The model is capable of differentiating between benign and malignant skin lesions and also provides a specific diagnosis for malignant cases. |
| |
|
| | ## Model Usage |
| | The model will be uploaded later this week in .bin format and .tar.gz. |
| |
|
| | You can use the SkinSense model by installing the `transformers` library from Hugging Face: |
| |
|
| |
|
| | ## Model Link |
| | You can access the pre-trained SkinSense model on Hugging Face Model Hub using the following link: |
| |
|
| | [SkinSense Model on Hugging Face]() |
| |
|
| | ## Model Training |
| |
|
| | If you're interested in the details of the model training process, you can find the code and instructions in the `model_training` directory of the [GitHub repository](https://github.com/Abdullah-Kazi/SkinSense). The training data, data augmentation techniques, and the ResNet101 architecture are used during the training process. The model's performance metrics, including accuracy and loss, are logged during training. |
| |
|
| | ## Model Evaluation |
| |
|
| | The performance of the model has been evaluated on a separate test dataset to assess its accuracy and other metrics. You can find the evaluation results in the `model_evaluation` directory of the [GitHub repository](https://github.com/Abdullah-Kazi/SkinSense). |
| |
|
| | ## Inference Speed and Hardware Requirements |
| |
|
| | The inference speed of the SkinSense model depends on the hardware used for prediction. On a GPU, the model can process multiple images simultaneously, significantly improving performance. For faster inference times, we recommend using a GPU with at least 8GB of VRAM. |
| |
|
| | ## Contributing |
| |
|
| | We welcome contributions to the SkinSense model. If you find any issues or want to enhance the model's performance, feel free to submit a pull request in the [GitHub repository](https://github.com/Abdullah-Kazi/SkinSense). Make sure to follow the code of conduct and provide clear documentation for your changes. |
| |
|
| | ## Contact |
| |
|
| | If you have any questions or inquiries related to the SkinSense model, you can reach out to: |
| |
|
| | - Abdullah Kazi |
| | - Phone: +1(925) 460-7273 |
| | - Email: kaziabdullah61@gmail.com, abdullah.kazi@mg.thedataincubator.com |
| | - LinkedIn: [Abdullah Kazi on LinkedIn](https://www.linkedin.com/in/abdullah1kazi/) |
| | - GitHub: [Abdullah-Kazi](https://github.com/Abdullah-Kazi) |
| | - HuggingFace: [Akazi](https://huggingface.co/Akazi) |
| |
|
| | ## License |
| | The SkinSense model is released under the MIT License. See the LICENSE file for more details. |
| |
|
