# DSMNet **Repository Path**: yrccry/DSMNet ## Basic Information - **Project Name**: DSMNet - **Description**: Height Prediction and Refinement from Aerial Images with Semantic and Geometric Guidance - **Primary Language**: Unknown - **License**: Not specified - **Default Branch**: main - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 0 - **Created**: 2020-12-14 - **Last Updated**: 2020-12-19 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README ## DSMNet ### About This repo contains the code and files necessary to reproduce the results published in our [paper](https://arxiv.org/pdf/2011.10697.pdf) 'Height Prediction and Refinement from Aerial Images with Semantic and Geometric Guidance'. Our method relies on a two stage pipeline : First, a multi-task network is used to predict the height, semantic labels and surface normals of an input RGB aerial image. Next, we use a denoising autoencoder to refine our height prediction in order to produce higher quality height maps. Training and testing is conducted on two publicly available datasets : The ISPRS Vaihingen and the IEEE DFC2018.  ### Citation If you find our work useful in your research, please consider citing: ``` @misc{mahdi2020height, title={Height Prediction and Refinement from Aerial Images with Semantic and Geometric Guidance}, author={Elhousni Mahdi and Huang Xinming and Zhang Ziming}, year={2020}, eprint={2011.10697}, archivePrefix={arXiv}, primaryClass={cs.CV} } ``` ## Network ### Prerequisites * Python 3.5 * Tensorflow 2.1 (with Cuda 10.0) * Numpy 1.18.4 ### Testing Both datasets can be found [here](https://drive.google.com/file/d/1llKA6z5L6CBQA5Fyr92d0Alj9TjhpYWb/view?usp=sharing). The data was organized and seperated into tiles to speed up the training process. No further pre-processing was done. Our checkpoints can be found [here](https://drive.google.com/file/d/1DkMmK1zvypZjqZ9GsIWqtdAO0ocEnTOQ/view?usp=sharing). When unzipping the datasets and checkpoints, make sure to respect the following folder structure : / -datasets --DFC2018 ---RGB ---SEM ---DSM ---DEM --Vaihingen ---RGB ---SEM ---NDSM -checkpoints --DFC2018 --Vaihingen Next, use the test_dsm.py script to test the height prediction and refinement networks by using : **python test_dsm.py [dataset] [refinement_flag]** For example, to test the results of the prediction and refinement networks combined on the DFC2018 dataset, use : python test_dsm.py DFC2018 True To test the results of the prediction network only on the Vaihingen dataset, use : python test_dsm.py Vaihingen False The output files will be saved to the /output folder. ### Training To train your own MTL prediction network, use: **python train_mtl.py [dataset]** For example, to train the MTL prediction network on the DFC2018 dataset, use : python train_mtl.py DFC2018