https://github.com/t2solve/realbugtracker
extract movement trajectory of moving objects in videos from image based tracking
https://github.com/t2solve/realbugtracker
Last synced: over 1 year ago
JSON representation
extract movement trajectory of moving objects in videos from image based tracking
- Host: GitHub
- URL: https://github.com/t2solve/realbugtracker
- Owner: t2solve
- License: mit
- Created: 2022-11-28T22:57:25.000Z (over 3 years ago)
- Default Branch: master
- Last Pushed: 2024-12-28T09:33:29.000Z (over 1 year ago)
- Last Synced: 2025-01-22T03:19:36.129Z (over 1 year ago)
- Language: C++
- Homepage:
- Size: 3.36 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# realbugtracker
cite this project with :
[](https://zenodo.org/badge/latestdoi/571799961)
(See gif shows an example of a moving Nebria brevicollis ( aka bug:) the red border shows the detected contour and the bigger red dot, the centroid of the moving contour)
## project aim
* extract movement trajectory of moving objects in videos from image based tracking
* process a small example (see download link) of a moving bug (species: Nebria brevicollis)
## supported os
* tested on linux debian / ubuntu builds
## license
MIT License
## dependency
* c++ library OpenCV (4.6 or above)
* c++ library bgs lib (3.2 or above)
* c libarchive (3.4 or above)
## needed input
* picture collection (frames) of a moving object (download example)
* time code of each frame in a text file (see example)
* neutral background
* fix orientation of camera to 2D arena surface
### preparation of the setup
1. install camera in top view position ( orthogonal and centered) of flat 2D surface e.g.
(the pictures shows a Prosilica GT 1920 camera from Allied Vision, which could be controlled via a c++ sdk vimba, not published here)
1. do a camera calibration, explained [here](https://docs.opencv.org/4.x/d4/d94/tutorial_camera_calibration.html) (use the code at the opencv repository [here](https://github.com/opencv/opencv/tree/4.x/samples/cpp/tutorial_code/calib3d/camera_calibration/camera_calibration.cpp) )
2. save the camera calibration file
# processing
3 software pieces are implemented for processing
# Tool 1: extractcenterpoints.cpp
aim: extract the moving objects, write polygons and center to one file for each frame, see the [code](src/extractcenterpoints.cpp) and read the steps
step 0) do lens correction, uses the camera calibration to make lens correction
step 1) cut the region of interest, to avoid any reflection artifacts
step 2) uses [bgslibrary](https://github.com/andrewssobral/bgslibrary) to extract the moving object, here we use method "Background Segmentation with Feedback: "The Pixel-Based Adaptive Segmenter" was used TODO cite paper here
step 3) apply edge detection
step 4) apply filter for contours sizes
step 5) calc hu moments and and mass center for the contour selection (optional, needed for visual debugging needed)
step 6) calc convex hull of all polygons
step 7) calc hu moments and and mass center (centroid) for the convex hull
step 8) write all contours, mass centers and the convex hull points to a yml file
# Tool 2: mergetimeline.cpp
aim: merge the polygon file and the timetable to on time line file of centroids, see the [code](src/mergetimeline.cpp)
step 0) read all files
step 1) select the centroid of the convex hull, if there is at least one polygon in the frame
step 2) write down the information to 1 file
# Tool 3: producetrajectory.cpp
aim: merge the polygon file and the timetable to on trajectory file
see the [code](src/producetrajectory.cpp)
step 0) project trajectory position data from camera space into real world space using the camera calibration file
step 1) summaries all position in a trajectory file
# compile the code
```bash
cmake .
make -j 10
cd bin
```
# process the example
## download example
dowload 316 mb example data from http://t2solve.com/realbugtracker/rec97371911.tar.gz to examples folder
or use the script [here](examples/downloadExample.sh) or use the following steps below. please verify the tar file via checksum
```bash
cd examples
echo "download the example file"
wget http://t2solve.com/realbugtracker/rec97371911.tar.gz
echo "download the checksum file "
wget http://t2solve.com/realbugtracker/rec97371911.tar.gz.sha512
echo "validate checksum"
sha512sum -c rec97371911.tar.gz.sha512
```
## unpack example
```bash
tar xzvf rec97371911.tar.gz
```
## read docu
```bash
#file meaning
more rec97371911/README.md
#howtp start Tool 1
more rec97371911/runStep1.sh
#howtp start Tool 2
more rec97371911/runStep2.sh
#howtp start Tool 3
more rec97371911/runStep3.sh
```
## run example
```bash
cd rec97371911
./run.sh
```
## hint 1: build bgslib on linux debian
```bash
git clone https://github.com/andrewssobral/bgslibrary
cd bgslibrary
git checkout v3.2.0 #use tested tagged version
cmake .
make -j 10
sudo make install
```