- 28.11.2013 - Framework source code available for download.
- 03.09.2013 - Dataset available for download.
High precision ground truth data is a very important factor for the development and evaluation of computer vision algorithms and especially for advanced driver assistance systems. Unfortunately, some types of data, like accurate optical flow and depth as well as pixel-wise semantic annotations are very difficult to obtain.
In order to address this problem, in this paper we present a new framework for the generation of high quality synthetic camera images, depth and optical flow maps and pixel-wise semantic annotations. The framework is based on a realistic driving simulator called VDrift (www.vdrift.net), which allows us to create traffic scenarios very similar to those in real life.
Synthetic Dataset VDrift from Vladimir Haltakov on Vimeo.
The code of the framework together with documentation and some examples is available on GitHub?
The images from the dataset used in the paper can be downloaded below. In order to interpret the depth and flow images we provide Matlab scripts which can be downloaded from here
* Camera images (468 MB)
* Depth images (401 MB)
* Flow images (585 MB)
* Pixelwise annotations (16 MB)
* Camera poses (0.3 MB)