Feat: Add to_csv.py(#2)

To_csv.py extracts train.csv, valid.csv by dividing the images by the specified ratio.

Tools/to_csv.py

@@ -0,0 +1,43 @@
+import pandas as pd
+import random
+from pathlib import Path
+
+############ Custom ############
+PORTION = 20  ## 몇퍼센트(백분율)를 validation으로 사용할 것인지
+SEED = 42
+IMG_DIR_PATH = '/opt/ml/sample_data/images'
+TRAIN_CSV_PATH = "./train.csv"
+VALID_CSV_PATH = "./valid.csv"
+############ Custom ############
+
+random.seed(SEED)
+train_list = []
+valid_list = []
+
+for actor in Path(IMG_DIR_PATH).iterdir():
+    if not actor.is_dir():
+        continue
+    image_path_list = []
+    count = 0
+    for image in actor.glob("*.jpg"):
+        count += 1
+        image_path_list.append(str(Path('').joinpath(*image.parts[-3: ])))
+
+    random.shuffle(image_path_list)
+
+    valid_num = count*PORTION // 100
+    train_path_list = sorted(image_path_list[valid_num:])
+    valid_path_list = sorted(image_path_list[:valid_num])
+
+    for image in train_path_list:
+        train_list.append([image, actor.name])
+
+    for image in valid_path_list:
+        valid_list.append([image, actor.name])
+
+
+df_train = pd.DataFrame(data=train_list, columns = ['path','name'])
+df_train.to_csv(TRAIN_CSV_PATH) ## train csv파일 위치 지정
+
+df_valid = pd.DataFrame(data=valid_list, columns = ['path','name'])
+df_valid.to_csv(VALID_CSV_PATH) ## valid csv파일이 저장될 위치 지정