Automatic Camera Calibration by Landmarks on Rigid Objects [under review]

Vojtěch Bartl, Jakub Špaňhel, Petr Dobeš, Roman Juránek, Adam Herout
GRAPH@FIT, Brno University of Technology
Corresponding authors: ibartl [at] fit.vutbr.cz

Abstract

This article presents a new method for automatic calibration of surveillance cameras. We are dealing with traffic surveillance and therefore the camera is calibrated by observing vehicles; however, other rigid objects can be used instead. The proposed method is using keypoints or landmarks automatically detected on the observed objects by a convolutional neural network. By using fine-grained recognition of the vehicles (calibration objects), and by knowing the 3D positions of the landmarks for the (very limited) set of known objects, the extracted keypoints are used for calibration of the camera, resulting in internal (focal length) and external (rotation, translation, scale) parameters of the surveillance camera. We collected a dataset in two parking lots and equipped it with a calibration ground truth by measuring multiple distances in the ground plane. This dataset seems to be more accurate than the existing comparable data (GT calibration error reduced from 4.62 % to 0.99 %). Also, the experiments show that our method overcomes the best existing alternative in terms of accuracy (error reduced from 7.71 % to 4.03 %) and our solution is also more flexible in terms of viewpoint change and other.

Downloads

  • Paper
  • Code – Implementation of camera calibration method in Python language with Tensorflow framework.

Citation

@article{Bartl2018_rigid,
author = "Vojtěch Bartl and Jakub Špaňhel and Petr Dobeš and Roman Juránek and Adam Herout",
title = "Automatic Camera Calibration by Landmarks on Rigid Objects",
journal = "Computer Vision and Image Understanding",
issn = "1077-3142",
}


Acknowledgment

This work was supported by The Ministry of Education, Youth and Sports of the Czech Republic from the National Programme of Sustainability (NPU II); project IT4Innovations excellence in science — LQ1602. Also, this work was supported by TACR project “SMARTCarPark”,
TH03010529.