By Iro Laina, Christian Rupprecht, Vasileios Belagiannis, Federico Tombari, Nassir Navab.
This repository contains the CNN models trained for depth prediction from a single RGB image, as described in the paper "Deeper Depth Prediction with Fully Convolutional Residual Networks". The provided models are those that were used to obtain the results reported in the paper on the benchmark datasets NYU Depth v2 and Make3D for indoor and outdoor scenes respectively. Moreover, the provided code can be used for inference on arbitrary images.
The trained models are currently provided in two frameworks, MatConvNet and TensorFlow. Please read below for more information on how to get started.
Prerequisites
The code provided in the matlab folder requires the MatConvNet toolbox for CNNs. It is required that a version of the library equal or newer than the 1.0-beta20 is successfully compiled either with or without GPU support.
Furthermore, the user should modify matconvnet_path = '../matconvnet-1.0-beta20' within evaluateNYU.m and evaluateMake3D.m so that it points to the correct path, where the library is stored.
How-to
For acquiring the predicted depth maps and evaluation on NYU or Make3D test sets, the user can simply run evaluateNYU.m or evaluateMake3D.m respectively. Please note that all required data and models will be then automatically downloaded (if they do not already exist) and no further user intervention is needed, except for setting the options opts and netOpts as preferred. Make sure that you have enough free disk space (up to 5 GB). The predictions will be eventually saved in a .mat file in the specified directory.
Alternatively, one could run DepthMapPrediction.m in order to manually use a trained model in test mode to predict the depth maps of arbitrary images.
The code provided in the tensorflow folder requires accordingly a successful installation of the TensorFlow library (any platform).
The model's graph is constructed in fcrn.py and the corresponding weights can be downloaded using the link below. The implementation is based on ethereon's Caffe-to-TensorFlow conversion tool.
predict.py provides sample code for using the network to predict the depth map of an input image.
The models are fully convolutional and use the residual learning idea also for upsampling CNN layers. Here we provide the fastest variant in which interleaving of feature maps is used for upsampling. For this reason, a custom layer +dagnn/Combine.m is provided.
The trained models - namely ResNet-UpProj in the paper - can also be downloaded here:
- NYU Depth v2: MatConvNet model, TensorFlow model
- Make3D: MatConvNet model, TensorFlow model (soon)
In the following tables, we report the results that should be obtained after evaluation and also compare to other (most recent) methods on depth prediction from a single image.
- Error metrics on NYU Depth v2:
| State of the art on NYU | rel | rms | log10 |
|---|---|---|---|
| Roy & Todorovic (CVPR 2016) | 0.187 | 0.744 | 0.078 |
| Eigen & Fergus (ICCV 2015) | 0.158 | 0.641 | - |
| Ours | 0.127 | 0.573 | 0.055 |
- Error metrics on Make3D:
| State of the art on Make3D | rel | rms | log10 |
|---|---|---|---|
| Liu et al. (CVPR 2015) | 0.314 | 8.60 | 0.119 |
| Li et al. (CVPR 2015) | 0.278 | 7.19 | 0.092 |
| Ours | 0.175 | 4.45 | 0.072 |
- Qualitative results:
If you use this method in your research, please cite:
@inproceedings{laina2016deeper,
title={Deeper depth prediction with fully convolutional residual networks},
author={Laina, Iro and Rupprecht, Christian and Belagiannis, Vasileios and Tombari, Federico and Navab, Nassir},
booktitle={3D Vision (3DV), 2016 Fourth International Conference on},
pages={239--248},
year={2016},
organization={IEEE}
}
Simplified BSD License
Copyright (c) 2016, Iro Laina
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
-
Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
-
Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.