Win the 2019 IEEE GRSS Data Fusion Contest – Large-Scale Semantic 3D Reconstruction. Track 3: Multi-view semantic stereo

This project is only for hackers, proficient with unix and python.

The 2019 IEEE GRSS Data Fusion Contest track 3 [2] concerns multi-view stereo:
Given multi-view images for each geographic tile, the objective is to predict semantic labels and a DSM. Unrectified images are provided with RPC metadata already adjusted using the lidar so that registration is not required in evaluation and so that solutions can focus on methods for image selection, correspondence, semantic labeling, and multi-view fusion. Since this track relies on RPC metadata which may not be familiar to everyone, the baseline algorithm provided includes simple python code to manipulate RPC for epipolar rectification and triangulation. Participants of Track 3 are intended to submit 2D semantic maps and DSMs in raster format (similar to the tif file of the training set). Performance is assessed using mIoU-3 with a threshold of 1 meter for the DSM Z values.

To win this challenge you will build on the expertise of the CMLA team on multi-date stereo reconstruction [1] and incorporate ML tools [3] for producing a the semantic segmentation of the scene along with the 3D reconstruction.

Supervision

Gabriele Facciolo, Carlo de Franchis, Enric Meinhardt

Bibliographic References

[1] G. Facciolo, C. de Franchis, and E. Meinhardt. Automatic 3D Reconstruction from Multi-Date Satellite Images., Earth Vision CVPRW, 2017.

[2] https://www.grss-ieee.org/community/technical-committees/data-fusion/data-fusion-contest/

[3] O. Ronneberger, P. Fischer, and T. Brox. U-net: Convolutional networks for biomedical image segmentation. MICCAI, 9351:234–241, 2015.

This entry was posted in Projets2019. Bookmark the permalink.