- Add seven more methods to the benchmark
- CR-Net
- SGPA
- iCaps
- SDFEst
- DPDN
- RBP-Pose
- Fix incomplete meshes in REAL275 dataset
- Fix mug offset for SPD-based methods
- Add NAD metric
- Cleaned up evaluation interface and code structure
CPASMethod
(defined incpas_toolbox/cpas_method.py
) defines the interface now, with all methods being inside thecpas_toolbox/cpas_methods
directory. - Add tutorial on how to run evaluation from another codebase