All notable changes to this project are documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
- [ci] added a first github CI for windows without cuda PR
- [ci] added more test images for linux PR
- compatibility issues with newer versions (>= 1.85) of boost PR
- Enable
-faligned-new
whenCCTAG_EIGEN_MEMORY_ALIGNMENT
is set on GCC >= 7.1 PR
- Replaced
CCTAG_EIGEN_MEMORY_ALIGNMENT
withCCTAG_EIGEN_MEMORY_ALIGNMENT
which has the opposite meaning. - Using the new TBB oneAPI interface, this breaks compatibility with previous versions of TBB PR
- Added documentation for conan PR
- Fixed potential out of bound in debug mode PR
- Renamed the marker files to be 0-based PR
- Fixed missing math module for boost PR
- Fixed compilation errors for newer version of boost on windows PR
- Removed old legacy defines for cuda and boost no more needed since the switch to c++14 PR
- Remove Cuda Cub dependency PR
- Bug fix: out of range image access during identification PR
- Bug fix on big markers (multi-resolution detection) PR
- Bug fix: avoid access to empty vector during vote PR
- CMake: export all symbols when building shared libs on Windows PR
- CMake: Improve management of CUDA Compute Capabilities flags PR
- Compatibility with Opencv 4 PR
- First Windows version (CPU only) PR
- Docker: add support for docker PR
- Improvements for ellipse fitting PR
- Modernizing code to C++11 PR
- Use Thrust for cuda >= 8 PR
- Minor code cleaning PR
- Add limit to edgepoint d-to-h transfer PR
- Bug fix in NearbyPoint PR
- Switch to modern version of CMake PR
- CVPR Publication https://hal.archives-ouvertes.fr/hal-01420665/document
- Application: Show detected markers in video and live mode PR
- Allow to extract CCTags from multiple images in parallel with multiple cuda streams PR PR
- Continuous integration on Travis PR
- Remove Blas/Lapack/Optpp dependencies
- Remove Ceres dependency
- GPU implementation
- First open-source release
- CPU Optimizations
- Expose critical parameters
- CMake build system
- Thesis Defence by Lilian Calvet under the direction of Vincent Charvillat and Pierre Gurdjos "Three-dimensional reconstruction methods integrating cyclic points: application to camera tracking" (http://www.theses.fr/2014INPT0002)
- CCTag detection on CPU