search

miendinh / VietnameseOCR

Vietnamese Optical Character Recognition. It works with Vietnamese and Latin characters as well.
Apache License 2.0
72 stars 40 forks source link

Checking custom image #5

Closed saurabhbidwai closed 6 years ago

saurabhbidwai commented 6 years ago

I want to check whether the model is working fine or not for the image beside this dataset for this I'm using below code but I'm getting a very bad prediction, Am I doing anything wrong?

import numpy as np from PIL import Image

img = Image.open('d.png').convert('RGBA')

new_width = 14 new_height = 14 img = img.resize((new_width, new_height), Image.ANTIALIAS)

arr = np.array(img)

flat_arr = arr.ravel()

type(flat_arr)

prob, idx = predict(flat_arr)

print('Predicted: ', characters[idx], ' with probability = ', prob, '%') print('-' * 10)

saurabhbidwai commented 6 years ago

Following code will help you to check custom images

import numpy as np from PIL import Image

img = Image.open('dd.png').convert('L')

new_width = 28 new_height = 28 img = img.resize((new_width, new_height), Image.ANTIALIAS)

arr = np.array(img, np.float32)

flat_arr = arr.ravel()

flat_arr = flat_arr.reshape((1,784))

prob, idx = predict(flat_arr)

print('Predicted: ', characters[idx], ' with probability = ', prob, '%') print('-' * 10)