ct_assist, or CameraTransform-Assistance, is a library for finding head-feet pairs on images. This allows for further automation of the CameraTransform. It extracts head-feet pairs from a binary mask generated using instance segmentation. This requires the user to input a pytorch tensor.
This project was made for EU-Project ASSISTANCE. See below for more.
This library will be applied to images of fluid spills in order to estimate their area and release rate. See fluid_estimator, request access from the the author.
This repo extends CameraTransform's functionality by trying to find the reference objects with Computer Vision.
We provide a Anaconda environment file [env.yaml] for installing this repo's dependencies.
git clone https://github.com/Casper-Smet/ct_assist.gitcd .\ct_assist\conda env create -f env.ymlconda activate ct_assistpip install .
Table: RMSE per camera property, ten runs
| n | Iterations | Roll degree | Tilt degree | Elevation in m |
|---|---|---|---|---|
| 1 | 1e4 | 56.23 | 20.91 | 1.02 |
| 2 | 1e4 | 54.97 | 22.40 | 1.01 |
| 3 | 1e4 | 47.68 | 18.59 | 1.00 |
| 4 | 1e4 | 52.01 | 23.76 | 1.01 |
| 5 | 1e4 | 50.09 | 21.32 | 0.99 |
| 6 | 1e4 | 48.51 | 20.06 | 0.98 |
| 7 | 1e4 | 51.90 | 21.73 | 0.98 |
| 8 | 1e4 | 47.26 | 22.09 | 0.95 |
| 9 | 1e4 | 50.51 | 22.29 | 1.01 |
| 10 | 1e4 | 51.74 | 20.88 | 0.96 |
| mean | 51.09 | 21.40 | 0.99 | |
| std | 2.94 | 1.42 | 0.02 |
Plot with results:
Adapted Situation AwareneSS tools and taIlored training scenarios for increaSing capabiliTies and enhANcing the proteCtion of First RespondErs – ASSISTANCE is an international research project funded by the European Commission under the Horizon 2020 programme in Secure Societies Challenge addressing the SU-DRS02-2018-2019-2020 (Technologies for first responders) topic.
The main purpose of ASSISTANCE project is twofold: to help and protect different kind of first responders’ (FR) organizations that work together during the mitigation of large disasters (natural or man-made) and to enhance their capabilities and skills for facing complex situations related to different types of incidents.
For more, see: https://assistance-project.eu/