remove bad images


#remove bad images
#encoding:utf-8 
## Used to find dirty pictures
import os
import re
import sys
from skimage import io,transform
from PIL import Image
# from PIL import ImageFile
# ImageFile.LOAD_TRUNCATED_IMAGES = True
print(1, "Warning:")
print(Image.MAX_IMAGE_PIXELS)

print(2, "No warning:")
Image.MAX_IMAGE_PIXELS = None
print(Image.MAX_IMAGE_PIXELS)

input_path = sys.argv[1]


remove_list = []
two_dimension = []
not_three_dimension = []
truncated_image = []
file_list = os.listdir(input_path)
for index,file_name in enumerate(file_list):
    file_path = os.path.join(input_path,file_name)
    print("is reading: ",index,  file_path)
    if re.findall(".jpg",file_path):
        try:
            img_file2 = io.imread(file_path)
            if (len(img_file2.shape) == 2):
                print("two dimension", file_name)
                print("two dimension", img_file2.shape)
                two_dimension.append(file_name)
            elif (len(img_file2.shape) != 3):
                print("not three dimension", file_name)
                print("not three dimension", img_file2.shape)
                not_three_dimension.append(file_name)
            elif (img_file2.shape[0]*img_file2.shape[1]*img_file2.shape[2] >= 178956970) or (img_file2.shape[2] > 3):
                print("images over pixels or not RGB", file_name)
                print("images over pixels or not RGB", img_file2.shape)
                remove_list.append(file_name)
        except Exception as e:
            truncated_image.append(file_name)
            print("image is truncated", file_name)
            print("image is truncated", img_file2.shape)
print("truncated_image", truncated_image)
print("two_dimension", two_dimension)
print("not_three_dimension", not_three_dimension)
print("remove_list", remove_list)

for remove_file_name in two_dimension :
    remove_file_path = os.path.join(input_path, remove_file_name)
    os.remove(remove_file_path)
    print("rm " + remove_file_path)
for remove_file_name in not_three_dimension:
    remove_file_path = os.path.join(input_path, remove_file_name)
    os.remove(remove_file_path)
    print("rm " + remove_file_path)
for remove_file_name in remove_list:
    remove_file_path = os.path.join(input_path, remove_file_name)
    os.remove(remove_file_path)
    print("rm " + remove_file_path)
for remove_file_name in truncated_image:
    remove_file_path = os.path.join(input_path, remove_file_name)
    os.remove(remove_file_path)
    print("rm " + remove_file_path)

你可能感兴趣的:(remove bad images)