DASZL: Dynamic Action Signatures for Zero-shot Learning

Authors: Tae Soo Kim, Jonathan Jones, Michael Peven, Zihao Xiao, Jin Bai, Yi Zhang, Weichao Qiu, Alan Yuille, Gregory D. Hager1817-1826

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

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate our method on the Olympic Sports and UCF101 datasets, where our model establishes a new state of the art under multiple experimental paradigms.
Researcher Affiliation Academia Tae Soo Kim*, Jonathan Jones* , Michael Peven , Zihao Xiao , Jin Bai , Yi Zhang , Weichao Qiu , Alan Yuille , Gregory D. Hager Johns Hopkins University 3400 N. Charles Street Baltimore, Maryland 21218 {tkim60, jdjones , mpeven, zxiao10, jbai12, yzhang286, wqiu7, ayuille1, hager}@jhu.edu
Pseudocode No The paper describes algorithms and uses diagrams to illustrate concepts, but it does not include a dedicated pseudocode block or clearly labeled algorithm steps formatted as code.
Open Source Code No The paper does not provide any explicit statement about releasing its source code or a link to a code repository for the methodology described.
Open Datasets Yes We evaluate our method on the Olympic Sports and UCF101 datasets, where our model establishes a new state of the art under multiple experimental paradigms. We also extend this method to form a unique framework for zero-shot joint segmentation and classification of activities in video and demonstrate the first results in zero-shot decoding of complex action sequences on a widely-used surgical dataset. Lastly, we show that we can use off-the-shelf object detectors to recognize activities in completely de-novo settings with no additional training. ... Olympic Sports (Niebles, Chen, and Fei-Fei 2010) and UCF101 (Soomro et al. 2012) datasets. ... JIGSAWS dataset (Gao et al. 2014) ... DIVA dataset1, which contains fine-grained human-object interactions under a real world video security footage. 1https://actev.nist.gov/
Dataset Splits No The paper describes training and test sets and their properties (e.g., 'randomly chosen seen/unseen classes (8/8 for Olympics, 51/50 for UCF101)'), but it does not explicitly provide details about a dedicated validation dataset split, its size, or its specific splitting methodology.
Hardware Specification Yes We gratefully acknowledge the support of NVIDIA Corporation with the donation of the Titan V GPUs used for this research.
Software Dependencies No The paper mentions using certain models or frameworks (e.g., 'pretrained TSM', 'off-the-shelf object detectors (He et al. 2017)') but does not provide specific version numbers for any software dependencies or libraries used in their implementation.
Experiment Setup No The paper states 'Optimization settings are provided in the supplementary material.', indicating that specific experimental setup details such as hyperparameters are not included in the main text of the paper.