This repository contains code to train the model for Skin Lesions Classification for benign and malignant classes, trained model and code to use the model. Sample Output:
We created a new dataset using the hair removal algorithm(uploaded). Interestingly, this method increased the accuracy of the model from 71% to 86%. You can run for more epochs and increase the accuracy significantly. I didnt upload the complete dataset, just few imagse to test the model. You can download and create one using the script I shared.
Origianl Dataset link: https://drive.google.com/file/d/134IIAM_GHQ3Dd6zI15HdBTD2ZTM5ATEs/
- Open the ModelLoading.py file
- Chagne the image path and run the code
- It'll load the model and predict on the given image
A few images I used from test set to test the model:
Class: Malignant
Class: Benign
Class: Malignant
Let me know in case of any issues or queries.
Thanks & Regards
TalhaPythoneer