CASA: Category-agnostic Skeletal Animal Reconstruction

Authors: Yuefan Wu, Zeyuan Chen, Shaowei Liu, Zhongzheng Ren, Shenlong Wang

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate CASA on both Planet Zoo and the real-world dataset DAVIS [46]. Experiments demonstrate that CASA recovers fine shape and realistic skeleton topology, handles a wide variety of animals, and adapts well to unseen categories. Additionally, we showed that CASA reconstructs a skeletal-animatable character readily compatible with downstream re-animation and simulation tasks.
Researcher Affiliation Academia Yuefan Wu1 Zeyuan Chen1 Shaowei Liu2 Zhongzheng Ren2 Shenlong Wang2 1 University of Science and Technology of China 2University of Illinois Urbana-Champaign
Pseudocode No The paper does not include any figure, block, or section explicitly labeled "Pseudocode" or "Algorithm".
Open Source Code No Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [No] we will release the code and data upon acceptance
Open Datasets Yes To this end, we propose a photo-realistic synthetic dataset Planet Zoo to study the dynamic animal reconstruction problem.
Dataset Splits Yes For Planet Zoo, we choose 24 out of 249 total animals for testing and use the rest for validation and training.
Hardware Specification Yes We use eight GTX 2080 Ti and RTX A6000 GPUs.
Software Dependencies No The paper mentions software tools like "Blender" and external libraries like "Point Rend [22]" and "volumetric correspondence net [73]" but does not provide specific version numbers for these or other software dependencies necessary for replication.
Experiment Setup No The paper states "We use Adam optimizer [21] to learn all the optimization variables. We adopt a scheduling strategy to avoid getting stuck at a local minimum." but does not provide specific experimental setup details such as learning rates, batch sizes, number of epochs, or other concrete hyperparameters in the main text.