A list of tools (python scripts) are created to make data processing easier and more convenient. Note that they are not professional tools so you may need to modify some lines before using it in your cases.

If you have more efficient tools, code or other suggestions to process DBNet data, especially point clouds, don't hesitate to contact @wangjksjtu(wangjksjtu@gmail.com) or submit pull-requests directly. Your contributions are highly encouraged and appreciated!

• img_pre.py: croping and resizing images using python-opencv
• las2fmap.py: extracting feature maps from point clouds
• pcd2las.py: downsampling point clouds; converting point clouds from '.pcd' to '.las' format.
• video2img.py: converting one video to continuous frames

To see HELP for these script:

python train.py -h

• python-opencv
• numpy, pickle, scipy, laspy
• CloudCompare (CC) (set PATH variables)

Convert point clouds to .las format:

CloudCompare.exe -SILENT -NO_TIMESTAMP -C_EXPORT_FMT LAS -O %s

Downsample point clouds to 16384 points and save in .las format:

CloudCompare.exe -SILENT -NO_TIMESTAMP -C_EXPORT_FMT LAS -O %s -SS RANDOM 16384

More command line usages of CloudCompareare available on the offical manual page.

Downlaod the example point cloud from the Google Drive.

python las2fmap.py -f example.las

To see HELP for the las2fmap.py script:

python las2fmap.py -h
# usage: las2fmap.py [-h] [-d DIR] [-f FILE]
# 
# optional arguments:
# -h, --help            show this help message and exit
# -d DIR, --dir DIR     Directory of las files [default: '']
# -f FILE, --file FILE  Specify one las file you want to convert # [default: '']