Selective and Orthogonal Feature Activation for Pedestrian Attribute Recognition

Authors: Junyi Wu, Yan Huang, Min Gao, Yuzhen Niu, Mingjing Yang, Zhipeng Gao, Jianqiang Zhao

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

Reproducibility Variable Result LLM Response
Research Type Experimental We conduct experiments on several benchmark PAR datasets, including PETA, PA100K, RAPv1, and RAPv2, demonstrating the effectiveness of our method. Specifically, our method outperforms existing state-of-the-art approaches by GRL, IAACaps, ALM, and SSC in terms of m A on the four datasets, respectively.
Researcher Affiliation Collaboration 1 AI Research Center, Xiamen Meiya Pico Information Company Ltd., Xiamen, China 2 Xiamen Meiya Pico Information Security Research Institute Company Ltd., Xiamen, China 3 College of Computer and Data Science, Fuzhou University, Fuzhou, China 4 Institute of Automation, Chinese Academy of Sciences, Beijing China 5 College of Physics and Information Engineering, Fuzhou University, Fuzhou, China
Pseudocode No The paper states: 'The detailed process of our SOFAFormer can be found in supplementary material.' This indicates the pseudocode or detailed process is not present in the main paper.
Open Source Code No The paper does not provide any explicit statements or links indicating the availability of open-source code for the described methodology.
Open Datasets Yes The experiments are conducted on four PAR benchmark datasets, including PETA (Deng et al. 2014), PA100K (Liu et al. 2017), RAPv1 (Li et al. 2016), and RAPv2 (Li et al. 2018b).
Dataset Splits No Details about these datasets and evaluation protocols can be found in the supplementary material. The main text does not explicitly provide the training/validation/test splits.
Hardware Specification No The paper does not explicitly describe the specific hardware (e.g., GPU/CPU models, memory) used for running its experiments.
Software Dependencies No The paper does not provide specific version numbers for any software dependencies used in the experiments.
Experiment Setup Yes We set α to 0.6 on the PA100K dataset to yield the best performance. We also use the m Five score to select the best parameters (i.e., τ = 0.6). The final loss of our SOFAFormer is the combination of Lbce and LOFA, which can be expressed as follows: L = Lbce + β LOFA (9) where β controls contribution of LOFA.