Social Image Parsing by Cross-Modal Data Refinement

Authors: Zhiwu Lu, Xin Gao, Songfang Huang, Liwei Wang, Ji-Rong Wen

IJCAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The experimental results on several benchmark datasets show the effectiveness of our algorithm.
Researcher Affiliation Collaboration 1School of Information, Renmin University of China, Beijing 100872, China 2CEMSE Division, KAUST, Thuwal, Jeddah 23955, Saudi Arabia 3IBM China Research Lab, Beijing, China 4School of EECS, Peking University, Beijing 100871, China
Pseudocode No The paper describes the algorithm steps in numbered text, but not in a formally structured pseudocode block or figure labeled "Algorithm" or "Pseudocode".
Open Source Code No The paper does not provide any specific links to source code or explicitly state that the code is publicly available.
Open Datasets Yes We select two benchmark datasets for performance evaluation: MSRC [Shotton et al., 2006] and Label Me [Liu et al., 2011]... we actually derive a Flickr dataset with realistic noise from the PASCAL VOC2007 benchmark dataset [Everingham et al., 2007].
Dataset Splits No It should be noted that the ground-truth pixel-level labels of all the images are unknown in our setting for image parsing. Hence, it is not possible to select the parameters by cross-validation for our CMDR algorithm.
Hardware Specification Yes We run all the algorithms (Matlab code) on a computer with 3.9GHz CPU and 32GB RAM.
Software Dependencies No The paper mentions "Matlab code" but does not provide specific version numbers for Matlab or any other software libraries or dependencies used.
Experiment Setup Yes In this paper, we thus uniformly set the parameters of our CMDR algorithm as k = 110, = 0.45 (equally λ = 0.82), and γ = 0.12 for the two benchmark datasets.