Submodular Attribute Selection for Action Recognition in Video

Authors: Jingjing Zheng, Zhuolin Jiang, Rama Chellappa, Jonathon P Phillips

NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on the Olympic Sports and UCF101 datasets demonstrate that the proposed attribute-based representation can significantly boost the performance of action recognition algorithms and outperform most recently proposed recognition approaches.
Researcher Affiliation Collaboration Jinging Zheng UMIACS, University of Maryland College Park, MD, USA zjngjng@umiacs.umd.edu; Zhuolin Jiang Noah s Ark Lab Huawei Technologies zhuolin.jiang@huawei.com; Rama Chellappa UMIACS, University of Maryland College Park, MD, USA rama@umiacs.umd.edu; P. Jonathon Phillips National Institute of Standards and Technology Gaithersburg, MD, USA jonathon.phillips@nist.gov
Pseudocode Yes Algorithm 1 Submodular Attribute Selection
Open Source Code No The paper does not contain any statement about releasing source code or a link to a code repository.
Open Datasets Yes In this section, we validate our method for action recognition on two public datasets: Sports dataset [25] and UCF101 [20] dataset.
Dataset Splits Yes Following the training and testing dataset partitions proposed in [30], we train a linear SVM and report classification accuracies of different attribute-based representations in Table 1.
Hardware Specification No The paper does not specify any hardware details such as GPU models, CPU types, or memory used for running the experiments.
Software Dependencies No The paper mentions using an SVM and other techniques like KSVD and PCA, but it does not provide specific version numbers for any software, libraries, or frameworks used.
Experiment Setup Yes Figure 4d shows the performance curves for a range of λ. We observe that the combination of entropy rate term and maximum coverage term obtains a higher classification accuracy than when only one of them is used. In addition, our approach is insensitive to the selection of λ. Hence we use λ = 0.1 throughout the experiments.