karlygzhu / RGAN

6 stars 0 forks source link

How to read the vid4 dataset? #1

Open Explore008 opened 11 months ago

Explore008 commented 11 months ago

3e22cc8e09f25567d2a09fd099bb310

karlygzhu commented 11 months ago

You need to put the four high-resolution sequences in the Vid4 dataset directly into the Vid4 folder, instead of adding an additional GT folder. For example, for 000.png in the calendar folder of the Vid4 dataset, the path to the image should be ./data_test/Vid4/calendar/000.png.

Explore008 commented 11 months ago

您需要将 Vid4 数据集中的四个高分辨率序列直接放入 Vid4 文件夹中,而不是添加额外的 GT 文件夹。例如,对于 Vid000 数据集的日历文件夹中的 4.png,图像的路径应为 ./data_test/Vid4/calendar/000.png。

Thank you for your timely reply! I have passed the vid4 test, and your work is excellent. Can you provide a test script that can be easily tested with other data sets, whether mp4 or png? thank you

karlygzhu commented 11 months ago

If you want to test other datasets, you need to place the video sequences from the other datasets in similar positions to Vid4. For example, for 0000.png in the archpeople folder of the UDM10 dataset, the path to the image should be ./data_test/UDM10/archpeople/0000.png. In addition, if frames in your video sequence are not in .png format, you need to modify the import type of frames to ensure that they can be imported smoothly. The modification location is line 27 of the test.py file, with the text inList = sorted(glob.glob(os.path.join(data_test, test_name, '*.png')))