| import os | |
| import pandas as pd | |
| DATA_CSV = "../../datasets/dfdc_test_preview/test_videos_preview_labels.csv" | |
| DATA_PATH = "../../datasets/dfdc_test_preview/faces" | |
| col_names = ["video", "label"] | |
| df_test = pd.read_csv(DATA_CSV, sep=' ', names=col_names) | |
| indexes_to_drop = [] | |
| for index, row in df_test.iterrows(): | |
| folders = os.listdir(os.path.join(DATA_PATH, row['video'])) | |
| if len(folders) < 2: | |
| indexes_to_drop.append(index) | |
| else: | |
| counter = 0 | |
| for folder in folders: | |
| if os.path.isdir(os.path.join(DATA_PATH, row['video'], folder)): | |
| counter += 1 | |
| if counter < 2: | |
| indexes_to_drop.append(index) | |
| df_test.drop(df_test.index[indexes_to_drop], inplace=True) | |
| df_test.to_csv("../../datasets/dfdc_test_preview/multi_identity_videos.csv") | |
| print(len(df_test)) | |