Closed CuttlefishXuan closed 7 months ago
You can download the image subdirectories from Dropbox or BaiduYun: https://www.dropbox.com/sh/8g2urrij2fercko/AABi0GclI-f96uYsAdP0D0Yga?dl=0 https://pan.baidu.com/s/1nqurXpbMzFo_-Cmf6eheOw?pwd=7zdg
You can download the image subdirectories from Dropbox or BaiduYun: https://www.dropbox.com/sh/8g2urrij2fercko/AABi0GclI-f96uYsAdP0D0Yga?dl=0 https://pan.baidu.com/s/1nqurXpbMzFo_-Cmf6eheOw?pwd=7zdg
Where are the original drone images in these directories?
image/ArtSci and image/PolyTech contain multiple sets of drone-captured images.
image/ArtSci and image/PolyTech contain multiple sets of drone-captured images.
How can I rebuild UrbanScene3D V1 with data under directory UrbanScene3D-dataset/images, especially Campus.zip? Because when I download Campus.zip? It's always invalid.
Maybe the loaddowned package is corrupted. You can also try to download it from https://pan.baidu.com/s/13OQej3m7rihxwUbGuY9ACQ?pwd=fb7j
@Linxius Sorry for the late reply and thanks for the BaiduYun links! It works! I have another question about the 3D segmentation data mentioned in the paper Sec.6. How can I acquire the single building models (e.g.meshs)? I expect to get the single building meshes and corresponding 2d images containing the building, format like the shapenet dataset, for the single-view building reconstruction task. I wonder if the UrbanScene3D dataset supports it.
We only provide building segmentations for the synthetic data in the simulator. You may try to obtain the segmentations by yourself, with pre-trained segmentation networks or something.
We only provide building segmentations for the synthetic data in the simulator. You may try to obtain the segmentations by yourself, with pre-trained segmentation networks or something.
okay, thank you
We only provide building segmentations for the synthetic data in the simulator. You may try to obtain the segmentations by yourself, with pre-trained segmentation networks or something.
Sorry I'm a little confused, how can I get the single building segmentation results and the labeled object id for the synthetic data? Cause I saw "we manually extract all single building models from the entire scene model" in this paper. Do you have the extracted single building models or do I have to process it myself?
The building models are only provided for certain scenes (see the figure above), which are contained in the simulator https://pan.baidu.com/s/1xkQxYA5XJJp2wIB1BAVVIg?pwd=wz5l
For other scenes like Campus, you need to process it by yourself.
Got it, I'll try to install the simulator.
Hi, is there a link to download the images individually?
The NAS link you provided is quite unstable, and the zip files I downloaded many times are invalid. I'm interested in the images in this dataset, can I just download the image data?