Coarse-to-fine Animal Pose and Shape Estimation

Authors: Chen Li, Gim Hee Lee

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We test our approach on the Stanford Extra dataset and achieve state-of-the-art results. Furthermore, we test the generalization capacity of our approach on the Animal Pose and BADJA datasets.
Researcher Affiliation Academia Chen Li Gim Hee Lee Department of Computer Science National University of Singapore {lic, gimhee.lee}@comp.nus.edu.sg
Pseudocode No The paper describes the proposed method in detail but does not include any explicit pseudocode or algorithm blocks.
Open Source Code Yes Our code is available at the project website1. 1https://github.com/chaneyddtt/Coarse-to-fine-3D-Animal
Open Datasets Yes We train our model on the Stanford Extra dataset [2], and test on the Stanford Extra, the Animal Pose [6] and the benchmark animal dataset of joint annotations (BADJA) [4] datasets.
Dataset Splits Yes Each dog-breed is divided into training and testing split with a ratio of 4 : 1.
Hardware Specification Yes The whole training process takes around 30 hours on a RTX2080Ti GPU.
Software Dependencies No Our model is implemented with Pytorch. A specific version number for Pytorch or any other software dependency is not provided.
Experiment Setup Yes We use the Adam optimizer with a learning rate of 10-4 in the first two stages and of 10-5 in the third stage.