Learning Patch-Based Dynamic Graph for Visual Tracking

Authors: Chenglong Li, Liang Lin, Wangmeng Zuo, Jin Tang

AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments show that our approach outperforms the state-of-the-art tracking methods on two standard benchmarks, i.e., OTB100 and NUS-PRO.
Researcher Affiliation Academia 1School of Computer Science and Technology, Anhui University, Hefei,China 2School of Data and Computer Science, Sun Yat-sen University, Guangzhou, China 3School of Computer Science and Technology, Harbin Institute of Technology, China
Pseudocode Yes Alg. 1 summarizes the optimization procedure.
Open Source Code No The paper does not provide any statement or link indicating that the source code for the described methodology is publicly available.
Open Datasets Yes We evaluate the proposed tracking method on the OTB100 benchmark dataset (Wu, Lim, and Yang 2015). [...] We also compare our approach with other tracking approaches on another large-scale benchmark dataset, NUS-PRO (Li et al. 2016a).
Dataset Splits No The paper mentions using OTB100 and NUS-PRO datasets for evaluation but does not specify explicit training, validation, and test splits with percentages, sample counts, or references to predefined splits for reproducibility.
Hardware Specification Yes The experiments are carried out on a PC with an Intel i7 4.0GHz CPU and 32GB RAM, and implemented in C++.
Software Dependencies No The paper states the implementation is "in C++" but does not provide specific version numbers for C++ compilers or any other libraries/dependencies used.
Experiment Setup Yes In Eq. (2), we empirically set {α, λ, β, γ, ξ} = {1, 0.1, 5, 18, 1}. In Struck, we empirically set {ω, θ} = {0.67, 0.2}. Besides, we partition all bounding box into 64 non-overlapping patches [...] each frame is scaled so that the minimum side length of a bounding box is 32 pixels, and the side length of a searching window is fixed to be 2 WH.