cshang412 / ground_normal_filter

Official implementation of: Towards Accurate Ground Plane Normal Estimation from Ego-Motion

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

ground_normal_filter

Official implementation of: Towards Accurate Ground Plane Normal Estimation from Ego-Motion
Paper | Arxiv | Youtube | Bilibili

TL;DR

Estimate ground plane normal vector from ONLY odometry information

Quick Start

📦 Prepare data

  • Download the KITTI Odometry dataset.
  • Prepare odometry results
    • One example has already been provided in odometry/orbslam2/00.txt.
    • You can download full results from here

🏗️️ Setup

python3 -m pip install -r requirements.txt  
sudo apt install ffmpeg  # for video visualization

🏃 Run the example

python3 src/run_kitti.py \
    --sequence 00 \
    --kitti_root /path/to/kitti/odometry/sequences \
    --pose_root odometry/orbslam2 \
    --output_root results
ffmpeg -y -framerate 10 -pattern_type glob -i 'results/vis/*.jpg' results/video.mp4

Citation

If you find our work useful, please consider citing our paper:

@article{zhang2022towards,
  title={Towards Accurate Ground Plane Normal Estimation from Ego-Motion},
  author={Zhang, Jiaxin and Sui, Wei and Zhang, Qian and Chen, Tao and Yang, Cong},
  journal={Sensors},
  volume={22},
  number={23},
  pages={9375},
  year={2022},
  publisher={Multidisciplinary Digital Publishing Institute}
}

About

Official implementation of: Towards Accurate Ground Plane Normal Estimation from Ego-Motion

License:MIT License


Languages

Language:Python 100.0%