This repository delivers EvLinearSolver, a linear solution for a series of geometric model fitting problems on event data.

[1] Motion and Structure from Event-based Normal Flow, Zhongyang Ren, Bangyan Liao, Delei Kong, Jinghang Li, Peidong Liu, Laurent Kneip, Guillermo Gallego, Yi Zhou, ECCV 2024. PDF, Video.

If you find this code useful in your research, please cite this work by:

@InProceedings{Ren2024eccv,
      title     = {Motion and Structure from Event-based Normal Flow}, 
      author    = {Zhongyang Ren and Bangyan Liao and Delei Kong and Jinghang Li and Peidong Liu 
                  and Laurent Kneip and Guillermo Gallego and Yi Zhou},
      booktitle = {European Conference on Computer Vision (ECCV)},
      doi       = {},
      year      = {2024}
      }

We have tested EvLinearSolver on machines with the following configurations

• Windows 10/11 + MATLAB R2022a

You may need add path to 'yamlmatlab' for hyperparameter settings

addpath(genpath('D:\Code\Mat\Utils')); % path to repo yamlmatlab

If you don't have this repo, please download it from https://github.com/ewiger/yamlmatlab

We have released calculated normal flow on serveral sequence which could be downloaded from https://drive.google.com/drive/folders/1-F3GzyCHIk9krUcTe2QiDxmnEYGnx8tb?usp=sharing.

• Rotational motion estimation: sequence shapes_rotation from ECD Dataset (IJRR, 2017).
• 6 DoF tracking estimation：sequence corner_slow1 from VECtor Dataset (RAL, 2021). More sequence will be made available soon. You can also calculate normal flow from your event data (Note that undistortion is necessary)

Given input sparse normal flow, our linear solver could give

To run this code, you need Clone this repository into your matlab workspace.

git clone https://github.com/NAIL-HNU/EvLinearSolver.git

You need to revised path in corresponding yaml file to run your code.

• Rotation Motion Estimation: You can run file rotation_test in the MATLAB_Code folder to check the result, this function output estimated angular velocity and groundtruth from IMU measurement.

• 6 DoF Tracking Estimation: You can run file tracking_test in the MATLAB_Code folder to check the result, this function output estimated angular velocity and linear velocity as well as groundtruth from IMU and Mocap System measurement.

• Differential Homography Estimation: will be updated soon.

If you have any questions or inquiries, please open an issue or email Zhongyang Ren at

• Primary: zhongyangren@hnu.edu.cn
• Alternate: cs.zhongyang@gmail.com (in case your email is filtered out by the primary mailbox).