Skip to content

sayedmohamedscu/skin-lesion-classifier-

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 

Repository files navigation

skin-lesion-classifier

classify skin lesions using Deep learning (convolutional neural network)

Getting Started

kaggle Dataset of skin lesions consists of 10015 images which describe 7 type of lesions.

Our goal is to build the best classifier and find the best technique to classify this lesions with high accurcy.

Or maybe getting ideas about building new hardware to get it more easier to be detected.

lets go

Prerequisites

you should have first 'any python IDE' Irecomend anacona or just work on colab if your gpu is not good.

You need keras machine learning library and other liberaries like numpy ,pandas and opencv which dealing with images and arrays and csv files

pip install Keras
pip install numpy
pip install pandas 
if you will work on your local pc (not recommended) 

Installing

If you will work on your local pc you need a lot of libraries but the most important one is the "pip"

It's built in in python 3.6 and higher

after setup python

open CMD

A fast way to launch this window is to press the Win + R keys on your keyboard. Then, type cmd and press Enter or click/tap OK

in cmd write this orders

pip install numpy
pip install opencv
pip install pandas
pip install keras

Running the tests for liberaries

check if you install it well

open python script then write

import cv2
import numpy 
import pandas
import keras

Results

unforchanitaly the 3 models give the same accurcy

due to

  • unbalance data

which mean each class doesn't have the same amount of data so the model becomes [Biased]

solutions will be add to this problem

  • the data itsalf

not clear and the normal camera'RGB' _i think it isn't suitable for scan like this , we need multispectral imaging _

Built With

Authors

License

This project is licensed under the MIT License - see the LICENSE.md file for details

Acknowledgments