Skin Cancer

In this project, we’ll learn to build an Image classification project for detecting skin cancer in the image using deep learning methods. Will learn to build CNN models and then improve the performance using Transfer learning approaches with the pre-trained models. machine learning technique that can be used to automatically detect cancerous lesions in images. This technique can be use to improve the accuracy of a skin cancer diagnosis.

a machine learning algorithm on a dataset of images, it can learn to distinguish between cancerous and non-cancerous skin lesions. There are many different image classification algorithms, but one of the most promising is convolutional neural networks (CNNs). CNN’s have been used to achieve state-of-the-art results in a variety of image classification tasks, including object detection and recognition.

What will you Learn in the Skin Cancer Image classification project?

1. Loading the image data with ImageDataGenerator class
2. Developing the CNN model and compiling and fitting it on the train set
3. How to perform Data Augmentation?
4. How to do transfer learning with pre-trained models such as ResNet50?

Prerequisite

1. Woking knowledge of Keras library
2. Understanding of different convolution methods and operations i.e. Filter, kernels, padding, pooling

Tools

1. Google Colab for training the model
2. Matplotlib library for visualizing the loss & accuracy of the model
3. Keras[tensorflow]

Tasks

We will be performing the following tasks in this project:

Task-1: Import libraries and load the data with ImageDataGenerator class

Task-2: Build a vanilla[base] CNN Model

Task-3: Compile and fit the base model

Task-4: Plot the results of loss and accuracy of the base model

Task-5: Perform different data augmentation techniques on the train set

Task-6: Build a new model on augmented data and perform model evaluation

Task-7: Transfer Learning with ResNet50

Task-8: Build a new ResNet Model and fit it on the dataset

Task-9: Inferencing on the test image with best performing model