The link to the preprocessed scannet200 data is available in the following link ScanNet200 preprocessed.
Important: please make sure to fill in the ScanNet terms of use, and send it to the scannet group email before downloading the pre-processed data.
Download ScaNnet200 and follow Mask3D to preprocess data. A preprocessed point cloud scene has the following format id.npy. Kindly move all preprocessed scenes to their corresponding folders e.g. 0011_00.npy to ./data/scannet200/scene0011_00.
Next, you need to download the RGB-D sequence, which is available in OpenScene repository, or from the official Scannet200 website ScaNnet200.
ScanNet200 data folder should be structured as follows
./data
βββ scannet200
βββ ground_truth
βββ scene0011_00.txt
βββ ...
βββ scene0011_00
βββ poses <- folder with camera poses
β βββ 0.txt
β βββ 1.txt
β βββ ...
βββ color <- folder with RGB images
β βββ 0.jpg (or .png/.jpeg)
β βββ 1.jpg (or .png/.jpeg)
β βββ ...
βββ depth <- folder with depth images
β βββ 0.png (or .jpg/.jpeg)
β βββ 1.png (or .jpg/.jpeg)
β βββ ...
βββ intrinsics.txt <- camera intrinsics
β
βββ 0011_00.npy <- preprocessed point cloud of the scene
β
βββ scene0011_00_vh_clean_2.ply <- raw point cloud of the scene
βββ ...
Run the following command.
sh scripts/get_replica_dataset.sh
A replica folder with the following format will be generated.
./data
βββ replica
βββ ground_truth
βββ office0.txt
βββ ...
βββ office0
βββ poses <- folder with camera poses
β βββ 0.txt
β βββ 1.txt
β βββ ...
βββ color <- folder with RGB images
β βββ 0.jpg (or .png/.jpeg)
β βββ 1.jpg (or .png/.jpeg)
β βββ ...
βββ depth <- folder with depth images
β βββ 0.png (or .jpg/.jpeg)
β βββ 1.png (or .jpg/.jpeg)
β βββ ...
βββ intrinsics.txt <- camera intrinsics
β
βββ office0_mesh.ply <- raw point cloud of the scene
βββ ...