Open kszpxxzmc opened 1 week ago
I would like to know if the downloaded image+pose data includes all the frames. Theoretically, a 60-second video at 60 FPS would have 3,600 images, but only about 300 images are provided. Whether you extract images with 60 FPS?
Hi @kszpxxzmc, we are processing the labels. 1~4K labels can be found in the preview page: https://htmlpreview.github.io/?https://github.com/DL3DV-10K/Dataset/blob/main/visualize/index.html.
As for the images, usually we do not extract frames with that high frame rates as 3600 4K PNG images are very huge and much bigger than a video. And also, it is hard for colmap to process so many images. We pick a practical number for image+pose, roughly 300 images per mins. If you need higher framerate, maybe you can download the 4K video and process them by yourself to satisfy customized needs.
Will keep you posted when all the labels are ready.
Thank you very much! Looking forward to your work!
I would like to confirm whether the dataset includes both the POI category and the secondary POI category as mentioned on your website.