vod-converter
vod-converter copied to clipboard
Convert between visual object detection datasets
Visual Object Dataset converter
Converts between object dataset formats.
Example: convert from data in KITTI format to Pascal VOC format:
$ python vod_converter/main.py --from kitti --from-path datasets/mydata-kitti --to voc --to-path datasets/mydata-voc
Usage
Flags:
-
--from-path
: Source dataset -
--from
: Type of source dataset. This flag only accepts 4 values:-
kitti
: KITTI. -
kitti-tracking
: KITTI tracking. -
voc
: Pascal VOC -
udacity-crowdai
: Udacity CrowdAI. -
udacity-autti
: Udacity AUTTI
-
-
--to-path
: Destination path -
--to
: Type of destination dataset. This flag only accepts 2 values:-
kitti
: KITTI. -
voc
: Pascal VOC
-
See main.py
for documentation on how to easily plug in additional data formats; you can define a function
that can read in your data into a common format, and it will be then ready to convert to any supported format.
Similarly, you can implement a single function that takes the common format and outputs to the filesystem in your format and you will be ready to convert from e.g VOC to yours.
Acknowledge
This repository is a fork from vod-converter but has some changes to be compatible with both Python 2 and 3