An Empirical Study of CLIP for Text-Based Person Search

Authors: Min Cao, Yang Bai, Ziyin Zeng, Mang Ye, Min Zhang

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

Reproducibility Variable Result LLM Response
Research Type Experimental This paper makes the first attempt to conduct a comprehensive empirical study of CLIP for TBPS and thus contribute a straightforward, incremental, yet strong TBPS-CLIP baseline to the TBPS community.
Researcher Affiliation Academia 1 School of Computer Science and Technology, Soochow University 2 School of Computer Science, Wuhan University 3 Harbin Institute of Technology, Shenzhen
Pseudocode No The paper describes methods and equations but does not include explicit pseudocode or algorithm blocks.
Open Source Code Yes The code is available at https://github.com/Flame-Chasers/TBPS-CLIP.
Open Datasets Yes Comparisons with other methods are carried out on three datasets: CUHK-PEDES (Li et al. 2017b), ICFG-PEDES (Ding et al. 2021), RSTPReid (Zhu et al. 2021).
Dataset Splits No The paper mentions "few-shot capabilities (5% training data)" but does not explicitly state the full training/validation/test dataset splits for its main experiments. Details are deferred to the Appendix, which is not part of the main text analysis.
Hardware Specification No No specific hardware details (GPU models, CPU types, or cloud platforms) are mentioned for the experiments.
Software Dependencies No The paper mentions "Py Torch" but does not specify its version or any other software dependencies with version numbers.
Experiment Setup Yes The paper discusses "Training Tricks" such as global gradients back-propagation, dropout, locking bottom layers, and soft label. It also mentions hyperparameters like "τs is a hyper-parameter and set to 0.1" and "training in just 5 epochs".