Unsupervised Action Segmentation via Fast Learning of Semantically Consistent Actoms

Authors: Zheng Xing, Weibing Zhao

AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our approach is evaluated on four benchmark datasets, and the results demonstrate that our method achieves state-of-the-art performance.
Researcher Affiliation Academia 1 Future Network of Intelligence Institute, School of Science and Engineering, The Chinese University of Hong Kong, Shenzhen, China 2 Guangdong Laboratory of Machine Perception and Intelligent Computing, Shenzhen MSU-BIT University, China {zhengxing, weibingzhao}@link.cuhk.edu.cn
Pseudocode Yes Algorithm 1: The split-and-merge (Sa M) algorithm.
Open Source Code Yes Code is available at https://github.com/y66y/SaM.
Open Datasets Yes Datasets. We assessed the efficacy of our approach on four benchmark datasets: Breakfast (BF)(Kuehne, Arslan, and Serre 2014), You Tube Instructional Videos (YTI) (Alayrac et al. 2016), Hollywood Extended (HE) (Bojanowski et al. 2014), and 50Salads (FS) (Stein and Mc Kenna 2013).
Dataset Splits No The paper mentions using benchmark datasets and discusses "split 1 on BF" for runtime comparison but does not provide specific train/validation/test dataset splits (percentages, sample counts, or explicit methodology for splitting) for reproducibility.
Hardware Specification No The paper mentions 'hours of GPU-intensive model training' but does not provide specific hardware details such as GPU/CPU models, memory, or processor types used for experiments.
Software Dependencies No No specific ancillary software details, such as library names with version numbers, were provided to replicate the experiment.
Experiment Setup Yes In the upcoming experimental section, we will conduct a detailed analysis of how the window size δ affects the performance of our algorithm. Furthermore, we will illustrate the robustness of our algorithm in response to changes in δ.