Automatic detection of repeated objects

This repository implements a method for automatic detection of repeated objects in images. The companion paper (and more about this method) can be found here.

Prerequisites

In order to quickly start using this code we advise you to install anaconda and follow this guide.

Creating a conda environment

conda create --name autosim python=3.6

source activate autosim

pip install --upgrade pip
pip install -r requirements.txt

Compiling the C++ library

mkdir -p build && cd build && cmake .. && make

Reproducing results from the companion paper

Recreate figure 3 from the companion paper:

python fig_prelabeling.py

Detect best three repeated objects:

python autosim.py -q coca.png -a im3_sub.png -l "build/libautosim.so" -m 0.8 -n 1000 -i 0 -r 4 -w 800

Args explanation:

• -q coca.png. Query image in which to detect repeated objects.
• -a im3_sub.png.png. A-contrario image for the a-contrario matching criterion compatible with repeated structures.
• -l build/libautosim.so. Path to the compiled libautosim library.
• -m 0.8. Matching threshold (between 0 and 1) for the a-contrario matching criterion.
• -n 1000. Maximum number of matches to be taken into account.
• -i 0. RANSAC information type: 0 - Typical RANSAC; 1 - RANSAC 2pts; 2 - RANSAC affine.
• -r 4. Rho as in rho-hyperdescriptors. It sets the threshold for considering several spatially close keypoints as a single keypoint.
• -w 800. It scales down the query image if its width is greater than 800.

Authors

• Mariano Rodríguez - web page
• Jean-Michel Morel
• Julie Delon - web page

License

The code is distributed under the permissive MIT License - see the LICENSE file for more details.

Uninstall

conda deactivate
conda-env remove -n autosim
rm -R /path/to/autosim

Github repository

https://github.com/rdguez-mariano/autosim