Running online SfM while image capturing!

https://yifeiyu225.github.io/on-the-flySfMv2.github.io/On_the_Fly%20SfMv2.mp4

You can either download one of the pre-built binaries(for windows only) or build the source code manually.

• Follow the instructions provided in the link below to complete the download and installation. document

The development environment is based on Windows 10 with Visual Studio Community 2022. The project is based on C++17, Qt 5.15.12, Python 3.10.11, PyTorch 1.13.1, CUDA 11.7, and cuDNN 8.8. It also incorporates a variety of open-source libraries, including Boost, Ceres Solver, OpenCV, FLANN, and FAISS. It is recommended to use vcpkg to compile and manage these libraries.

• Checkout the latest source code:

git clone https://github.com/sygant/onthefly.git

• Download the pretrained weights for image retrival, and then copy the files to the following folder:
  /src/Feature/GlobalFeature/

• Download the Python Environment, and then copy the folder to the following path:
  /thirdparty/, example file path is like: /thirdparty/Python/python.exe

• Modify the corresponding Python environment path in the PyLoader class.

If you use this project for your research, please cite:

@article{zhan2025sfm,
  title={SfM on-the-fly: A robust near real-time SfM for spatiotemporally disordered high-resolution imagery from multiple agents},
  author={Zhan, Zongqian and Yu, Yifei and Xia, Rui and Gan, Wentian and Xie, Hong and Perda, Giulio and Morelli, Luca and Remondino, Fabio and Wang, Xin},
  journal={ISPRS Journal of Photogrammetry and Remote Sensing},
  volume={224},
  pages={202--221},
  year={2025},
  publisher={Elsevier}
} 

This work was jointly supported by the National Natural Science Foundation of China (Grant No. 42301507, 61871295), Natural Science Foundation of Hubei Province, China (No. 2022CFB727) and ISPRS Initiatives 2023.