CREStereo Repository for the 'Towards accurate and robust depth estimation' project
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Cpt.Captain 2731ef1ada changed some stuff, re-added default CREStereo scheduler 2 years ago
cfgs changed some stuff, re-added default CREStereo scheduler 2 years ago
doc/img Fix implementation issues 3 years ago
frontend frontend/__init__.py: general improvements, but still kinda wip 2 years ago
function_convertion_tests Initial commit 3 years ago
models Removed model 3 years ago
nets fix lightning, prepare sweeps 2 years ago
.gitattributes Initial commit 3 years ago
.gitignore enable training 3 years ago
README.md Removed model 3 years ago
api_server.py change a bunch of stuff, add wip lightning implementation 2 years ago
convert_to_onnx.py Fix flow_width 3 years ago
convert_weights.py #1 Added weight conversion to Pytorch 3 years ago
dataset.py changed some stuff, re-added default CREStereo scheduler 2 years ago
test_model.py change a bunch of stuff, add wip lightning implementation 2 years ago
train.py fix lightning, prepare sweeps 2 years ago
train_lightning.py changed some stuff, re-added default CREStereo scheduler 2 years ago

README.md

CREStereo-Pytorch

Non-official Pytorch implementation of the CREStereo (CVPR 2022 Oral) model converted from the original MegEngine implementation.

!CREStereo-Pytorch stereo detph estimation

Important

  • This is just an effort to try to implement the CREStereo model into Pytorch from MegEngine due to the issues of the framework to convert to other formats (https://github.com/megvii-research/CREStereo/issues/3).
  • I am not the author of the paper, and I am don't fully understand what the model is doing. Therefore, there might be small differences with the original model that might impact the performance.
  • I have not added any license, since the repository uses code from different repositories. Check the License section below for more detail.

Pretrained model

  • Download the model from here and save it into the models folder.
  • The model was covnerted from the original MegEngine weights using the convert_weights.py script. Place the MegEngine weights (crestereo_eth3d.mge) file into the models folder before the conversion.

Licences:

References: