3D-PhysNet: Learning the Intuitive Physics of Non-Rigid Object Deformations

Authors: Zhihua Wang, Stefano Rosa, Bo Yang, Sen Wang, Niki Trigoni, Andrew Markham

IJCAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results show the viability and the generalisation properties of the proposed architecture. ... 4 Experiments ... 4.3 Results and Discussion
Researcher Affiliation Academia Zhihua Wang 1, Stefano Rosa 1, Bo Yang1, Sen Wang2, Niki Trigoni1 and Andrew Markham1 1Department of Computer Science, University of Oxford 2School of Engineering and Physical Sciences, Heriot-Watt University {name.surname}@cs.ox.ac.uk, s.wang@hw.ac.uk
Pseudocode No The paper describes the network architecture and implementation details (Table 1) but does not include any explicitly labeled 'Pseudocode' or 'Algorithm' block.
Open Source Code No The paper states, 'The full generated datasets will be released for the sake of reproducibility and testing.', but it does not mention releasing the source code for the methodology itself.
Open Datasets Yes The full generated datasets will be released for the sake of reproducibility and testing.
Dataset Splits No The paper mentions 'training data' and 'test on unseen values' but does not specify exact training/validation/test dataset splits by percentages or counts.
Hardware Specification Yes The network was implemented with Tensorflow 1.4 and trained on a single Nvidia Pascal Titan X GPU.
Software Dependencies Yes The network was implemented with Tensorflow 1.4
Experiment Setup Yes The network was trained with a batch size of 8 using the Adam optimizer, with lr = 5e 5, β1 = 0.5, β2 = 0.999,ϵ = 1e 8. λ is set to 10 in Eq. 8.