Saliency Transfer: An Example-Based Method for Salient Object Detection

Authors: Xin Li, Fan Yang, Leiting Chen, Hongbin Cai

IJCAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Qualitatively and quantitatively experiments on six popular benchmark datasets validate that our approach greatly outperforms the state-of-the-art algorithms and recently published works.
Researcher Affiliation Academia 1School of Computer Science & Engineering, University of Electronic Science and Technology of China 2Institute of Electronic & Information Engineering in Dongguan, UESTC 3Digital Media Technology Key Laboratory of Sichuan Province
Pseudocode No The paper describes the steps of the proposed method in textual paragraphs but does not include any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not contain an explicit statement about releasing the source code for their proposed method, nor does it provide a link to a code repository. It mentions using 'the code or the saliency maps published by the authors of each method' which refers to comparison methods, not their own.
Open Datasets Yes To evaluate the proposed saliency transfer approach (abbreviated to STR ), standard benchmark datasets, MSRA1000 [Liu et al., 2011b], ECSSD [Yan et al., 2013], SED1 [Alpert et al., 2012], SED2 [Alpert et al., 2012], SOD [Movahedi and Elder, 2010] and i Co Seg [Batra et al., 2009], are used.
Dataset Splits No The paper mentions using several standard benchmark datasets for evaluation, but it does not provide specific details regarding training, validation, and test dataset splits (e.g., percentages, sample counts, or citations to predefined splits) that are explicitly used for its own experiments.
Hardware Specification Yes The average running time of each method is tested on a PC with an i5 2.50 GHz CPU and 8GB RAM
Software Dependencies No The paper states that 'Our STR is implemented by using MATLAB' but does not provide specific version numbers for MATLAB or any other software dependencies.
Experiment Setup Yes Input images are resized to be 400 x 300 pixels or 300 x 400 pixels beforehand... We set = 0.0005, β = 1, and d = 40 in Formula 5. We set µ = 0.3 and ' = 0.5 in Formula 6. The reference set includes 14 images carefully chosen from three publicly available datasets. These parameters and references are fixed in the following experiments.