Robustifying Generalizable Implicit Shape Networks with a Tunable Non-Parametric Model

Authors: Amine Ouasfi, Adnane Boukhayma

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the improvement obtained through our method with respect to baselines and the state-of-the-art using synthetic and real data.
Researcher Affiliation Academia Amine Ouasfi Adnane Boukhayma Inria, Univ. Rennes, CNRS, IRISA, M2S, France
Pseudocode Yes Algorithm 1 The training procedure of our method.
Open Source Code No The paper does not include an unambiguous statement that the authors are releasing the code for the work described in this paper, nor does it provide a direct link to a source-code repository for their method.
Open Datasets Yes Shape Net [12] consists of various instances of 13 different synthetic 3D object classes. Scan Net v2[18] is a challenging real dataset containing 1513 room scans as captured with an RBG-D camera. Faust [6] consists of real scans of 10 human body identities in 10 different poses.
Dataset Splits No The paper mentions using a 'training split' and 'test split' of Shape Net, but does not provide specific percentages, sample counts, or explicit citations to predefined standard splits for reproduction.
Hardware Specification Yes Our method takes roughly 10 seconds to converge on a NVIDIA RTX A6000 GPU.
Software Dependencies No The paper mentions using the 'Falcon Library' but does not specify its version number, nor does it list other software dependencies with specific version numbers.
Experiment Setup Yes We run all the experiments for 100 epochs using Adam optimizer with learning rate 0.1. We experimented with NP = 10k (also 500 and 3k in ablation) and NF = 32 and m = 500 inducing points.