FLIC: Fast Linear Iterative Clustering With Active Search

Authors: Jiaxing Zhao, Bo Ren, Qibin Hou, Ming-Ming Cheng, Paul Rosin

AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive evaluations on the Berkeley segmentation benchmark verify that our method outperforms competing methods under various evaluation metrics. Our method is implemented in C++ and runs on a PC with an Intel Core i7-4790K CPU with 4.0GHz, 32GB RAM, and 64 bit operating system. We compare our method with many previous and current state-of-the-art works... on the BSDS500 benchmark, using the evaluation methods proposed in (Arbelaez et al. 2011; Stutz, Hermans, and Leibe 2014).
Researcher Affiliation Academia Jiaxing Zhao, Bo Ren Nankai University Qibin Hou, Ming-Ming Cheng Nankai University Paul L. Rosin Cardiff University
Pseudocode Yes Algorithm 1 FLIC
Open Source Code No To facilitate the development of over-segmentation, the code will be publicly available.
Open Datasets Yes We compare our method with many previous and current state-of-the-art works, including FH (Felzenszwalb and Huttenlocher 2004), SLIC (Achanta et al. 2012), Manifold SLIC (Liu et al. 2016), SEEDS (Van den Bergh et al. 2012), and ERS (Liu et al. 2011) on the BSDS500 benchmark, using the evaluation methods proposed in (Arbelaez et al. 2011; Stutz, Hermans, and Leibe 2014). As in previous research in the literature (Liu et al. 2016; Wang et al. 2013a), we evaluate all algorithms on 200 randomly selected images of resolution 481 321 from the Berkeley dataset.
Dataset Splits No The paper states they evaluate on 200 randomly selected images from the Berkeley dataset, but does not specify how these images were split into training, validation, and test sets, or if standard benchmark splits were used for these purposes.
Hardware Specification Yes Our method is implemented in C++ and runs on a PC with an Intel Core i7-4790K CPU with 4.0GHz, 32GB RAM, and 64 bit operating system.
Software Dependencies No The paper states 'Our method is implemented in C++' but does not provide specific version numbers for any software libraries, frameworks, or other dependencies used in the experiments.
Experiment Setup Yes In our approach, three parameters need to be set. The first one is the number of superpixels K. ... The second one is the spatial distance weight m. ... The last parameter is the number of iterations itr. Here we set itr = 2 in default to get the balance between time cost and performance. ... we set m = 5 as default.