MSCNNS (Multi-scale Sub-pixel Convolutional Network with a Neighborhood Smoothness constraint) is a CNN-based approach for monocular depth estimation.
For technical details, please see this paper (comming soon).
- Matlab R2017a (or other proper version)
- python v3.5.x
- pytorch v0.3.0 (or later version)
- numpy
- scipy
You may use the provided model (see the BaiduYun link below) and test samples to test this apporach as follows,
python3 test.py --model <the pytorch model> --image ./test_samples/nyu_v2_175.mat
- Download The Dataset and The Train/Test Split file.
- Suppose you have saved the dataset in <path_to_data> and the split file in <path_to_split>. Open
matlab/gen_test_data_for_mscn.m
and assign <path_to_data> to 'NYUv2_data' and <path_to_split> to 'split_file'. - run
matlab/gen_test_data_for_mscn.m
and the test data will be generated in '../Dataset/test'. You may change the save root 'test_root' to anywhere you like. - Download the model in the
BaiduYun disk (Link: https://pan.baidu.com/s/1U0hw58K2M0y5QE4c3hbNng password: qnv3)
- Test the model as follows,
python3 test.py --model <the pytorch model> --data <folder for the generated test data>
Note that you may find the references and more comparisons in the aforementioned paper.