This is the official repository of the OMNICV-CVPR2022 workshop named "Pose Estimation for Two-View Panoramas: a Comparative Analysis"
Visit our paper:
- Create a conda enviorenment
conda create --name py37-keypoints python=3.7
conda activate py37-keypoints
-
Install the Spherical Package from Meder github
-
Git clone the project and enter the folder
git clone https://github.com/Artcs1/Keypoints.git
cd Keypoints
- Install the requirements
pip install -r requirements.txt
-
Install liegroups package in utils directory
-
Compile the SPHORB package
cd SPHORB-master
conda create --name SPHORB python=3.7
conda activate SPHORB
conda install -c conda-forge opencv==3.4.2
mkdir build1
cd build1
cmake ..
make
conda deactivate
- Compile the five-point algorithm package
cd fivepoint
conda create --name FP python=3.7
conda activate FP
conda install -c conda-forge opencv==4.5.5
mkdir build
cd build
cmake ..
make
conda deactivate
- Activate the conda enviorenment
conda activate py37-keypoints
- Run extrack_keypoints.py file
python3 extract_keypoints.py --datas [Urban1|Urban2|Urban3|Urban4|Room|Classroom|Realistic|Interior1|Interior2] --descriptors [sift|tsift|orb|torb|spoint|tspoint|sphorb]
- Example for Outdoor dataset
python3 extract_keypoints.py --datas Urban1 Urban2 Urban3 Urban4 --descriptors orb torb sift tsift spoint tspoint sphorb
- Example for Indoor dataset
python3 extract_keypoints.py --datas Room Classroom Realistic Interior1 Interior2 --descriptors orb torb sift tsift spoint tspoint sphorb
python3 show_keypoints.py --descriptor sift
@InProceedings{Murrugarra-Llerena_2022_CVPR,
author = {Murrugarra-Llerena, Jeffri and da Silveira, Thiago L. T. and Jung, Claudio R.},
title = {Pose Estimation for Two-View Panoramas Based on Keypoint Matching: A Comparative Study and Critical Analysis},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
month = {June},
year = {2022},
pages = {5202-5211}
}