dataset_automate_1.6.py

import cv2
import matplotlib.pyplot as plt
import numpy as np
from PIL import Image, ImageFilter, ImageEnhance
import pytesseract as pt
import os

# load the bottle_cascade
bottle_cascade=cv2.CascadeClassifier(/path1)#path of the cascade

img=cv2.imread(/path2)#path of the image

#preprocess the image to fit the 500*500 size specified for the classifier
img=cv2.resize(img,dsize=(500,500))
img_gray=cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)

#crop_dims contains the dimentsions of all the bottles in the format(x,y)the inital point and (w,h)width and height
crop_dims=bottle_cascade.detectMultiScale(img_gray,1.112,2)

crop_imgs=[]
for (x,y,w,h) in crop_imgs:
    cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)
    roi = img[y:y+h, x:x+w]
    crop_imgs.append(img[y:y+h, x:x+w])
#crop_imgs contains all the cropped images

#detecting the label text
crop_imgs_prepro=[]
img_labels=[]
for i in range(len(crop_imgs)):
    img=Image.fromarray(crop_imgs[i])
    img = img.convert('L')
    img = img.filter(ImageFilter.MedianFilter())
    enhancer = ImageEnhance.Contrast(img)
    img = enhancer.enhance(2)
    img = img.convert('1')
    crop_imgs_prepro.append(img)
    imagetext = pt.image_to_string(img)
    img_labels.append(imagetext)

#img_labels contain all the labels of the images

for i in range(len(crop_imgs)):
    path = ('path/',folderpath,'/',img_labels[i]) # folderpath is one of(train,test,val) which would be taken from the user
    #saving the cropped images in folders
    cv2.imwrite(os.path.join(path , str(i), crop_imgs[i])