diff --git a/scripts/dataset_validation.py b/scripts/dataset_validation.py index d660ce5..7250c24 100644 --- a/scripts/dataset_validation.py +++ b/scripts/dataset_validation.py @@ -22,14 +22,23 @@ def check_images(directory_path: str): except Exception: print(file) -def check_labels(dir_path: str) -> int: +def check_labels(dir_path: str): for filename in os.listdir(dir_path): name, extension = filename.split(".") if len(name) != 17: print(filename) - + +def check_symbols(dir_path: str): + with open(os.path.join(dir_path, "dict.txt"), "r") as dict_file: + dict_content = dict_file.readlines() + dict_chars = set(char.strip() for char in dict_content) + for filename in os.listdir(os.path.join(dir_path, "images")): + label, extension = filename.split(".") + if any([char not in dict_chars for char in label]): + print(filename) # rename_files("train_data/images/") # check_images("train_data/images/") -check_labels("train_data/images/") +# check_labels("train_data/images/") +check_symbols("train_data/")