dataset_man.py

from turtle import pos
import numpy as np
import pandas as pd
import os
import os.path
import shutil
from pathlib import Path
import glob


folder1 = 'main_dataset' #Read the label

metadata = pd.read_csv(os.path.join(folder1, 'metadata.csv')) #metadata.csv contains the dataset information
labels = pd.read_csv(r"C:\Users\parib\Desktop\COVID-19-PROJ\main_dataset\label.csv ", names=['label','condition']) #condition
true_label = labels[labels['condition'] == 'TRUE'] #covid pos
false_label = labels[labels['condition']== 'FALSE'] #covid neg
positive_labels = true_label.label
negative_labels = false_label.label
positive_list = []
negative_list = []

for i in positive_labels: #adding each positive and negative to their dedicated folder
    positive_list.append(i)

for i in negative_labels:
    negative_list.append(i)

label = labels['label'] #get the label, subject ID and file location
#condition = labels['condition']
subjectid = metadata['Subject ID'] #this is like the primary tag to find each img
filelocation = metadata['File Location'] #address of each img

filelocation_corrected = []

for i in filelocation: #address modification to capture img
    address = (i[17:])
    replace = address.replace('/','\\')
    filelocation_corrected.append(replace)


pos_dest = "Positive"
source = "covidimagesoutputPNGonly"

files = os.listdir(source)

src = "covidimagesoutputPNGonly"
dest = "Negative"


def assign_category(): #moving between folders
    for each in negative_list:
        try:
            shutil.copytree(src + "//" + each, dest + "//" + each)
        except FileNotFoundError:
            print ("File: ", each, "not found")


positive_folders = "Positive"
Negative_folders = "Negative"

def pos_png_only(): #seperating out the positive and negative
    files = glob.glob("./Positive/**/*.PNG", recursive = True)
    return files

def neg_png_only():
    files = glob.glob("./Negative/**/*.PNG", recursive = True)
    return files

pos_pics_destination = "Positivepics"
neg_pics_destination = "Negativepics"

if __name__ == "__main__":
    pos = pos_png_only()
    neg = neg_png_only()

    for each in pos:
        shutil.copy(each,pos_pics_destination) #copying each category
    print("Done copying pos pics")

    for  each in neg:
        shutil.copy(each, neg_pics_destination)
    print("Done copying neg pics")
    
    count = 0
    count1 = 0

    for i in pos:
        count += 1

    for i in neg:
        count1 += 1

    print("Positive pics = ", count, "Negative pics = ", count1) #making sure we have the right amout of files