Real-time 3D semi-local surface patch extraction using GPGPU

Feature vectors can be anything from simple surface normals to more complex feature descriptors. Feature extraction is important to solve various computer vision problems: e.g. registration, object recognition and scene understanding. Most of these techniques cannot be computed online due to their complexity and the context where they are applied. Therefore, computing these features in real-time for many points in the scene is impossible. In this work, a hardware-based implementation of 3D feature extraction and 3D object recognition is proposed to accelerate these methods and therefore the entire pipeline of RGBD based computer vision systems where such features are typically used. The use of a GPU as a general purpose processor can achieve considerable speed-ups compared with a CPU implementation. In this work, advantageous results are obtained using the GPU to accelerate the computation of a 3D descriptor based on the calculation of 3D semi-local surface patches of partial views. This allows descriptor computation at several points of a scene in real-time. Benefits of the accelerated descriptor have been demonstrated in object recognition tasks. Source code will be made publicly available as contribution to the Open Source Point Cloud Library.

Non-parametrized functionals with empirical dispersion corrections: A happy match?

The performances of two parametrized functionals (namely B3LYP and B2PYLP) have been compared with those of two non-parametrized functionals (PBE0 and PBE0-DH) on a relatively large benchmark set when three different types of dispersion corrections are applied [namely the D2, D3 and D3(BJ) models]. Globally, the MAD computed using non-parametrized functionals decreases when adding dispersion terms although the accuracy not necessarily increases with the complexity of the model of dispersion correction used. In particular, the D2 correction is found to improve the performances of both PBE0 and PBE0-DH, while no systematic improvement is observed going from D2 to D3 or D3(BJ) corrections. Indeed when including dispersion, the number of sets for which PBE0-DH is the best performing functional decreases at the benefit of B2PLYP. Overall, our results clearly show that inclusion of dispersion corrections is more beneficial to parametrized double-hybrid functionals than to non-parametrized ones. The same conclusions globally hold for the corresponding global hybrids, showing that the marriage between non-parametrized functionals and empirical corrections may be a difficult deal.