diff --git a/scripts/dataset_validation.py b/scripts/dataset_validation.py index 7250c24..7cc6a9e 100644 --- a/scripts/dataset_validation.py +++ b/scripts/dataset_validation.py @@ -1,5 +1,6 @@ import os import cv2 +from PIL import Image def rename_files(directory_path: str): extensions = set() @@ -37,8 +38,29 @@ def check_symbols(dir_path: str): if any([char not in dict_chars for char in label]): print(filename) +def max_height(dir_path: str): + max_height = 0 + max_filename = '' + for filename in os.listdir(dir_path): + im = Image.open(os.path.join(dir_path, filename)) + if im.height > max_height: + max_height = im.height + max_filename = filename + + print(max_filename, max_height) + +def resize_to_height(dir_path: str, target_height=48): + for filename in os.listdir(dir_path): + with Image.open(os.path.join(dir_path, filename)) as img: + width_percent = target_height / float(img.height) + new_width = int(float(img.width) * width_percent) + resized_img = img.resize((new_width, target_height), Image.LANZOS) + resized_img.save(os.path.join(dir_path, filename)) + + + # rename_files("train_data/images/") # check_images("train_data/images/") # check_labels("train_data/images/") -check_symbols("train_data/") - +# check_symbols("train_data/") +# max_height("train_data/images")