Joint M-Best-Diverse Labelings as a Parametric Submodular Minimization
Authors: Alexander Kirillov, Alexander Shekhovtsov, Carsten Rother, Bogdan Savchynskyy
NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | These theoretical results suggest two simple yet efficient algorithms for the joint M-best diverse problem, which outperform competitors in terms of runtime and quality of results. ... In particular, as we show in the paper, the new methods compute the exact M-best diverse labelings faster than a popular method of Batra et al., which in some sense only obtains approximate solutions. ... We describe and experimentally evaluate the two simplest of them, sequential and parallel. Both are considerably faster than the popular technique [4] and are as easy to implement. We demonstrate the effectiveness of these algorithms on two publicly available datasets. ... 5 Experimental Evaluation We base our experiments on two datasets: (i) The interactive foreground/background image segmentation dataset utilized in several papers [4, 31, 22, 23] for comparing diversity techniques; (ii) A new dataset for foreground/background image segmentation with binary pairwise energies derived from the well-known PASCAL VOC 2012 dataset [11]. |
| Researcher Affiliation | Academia | Alexander Kirillov1 Alexander Shekhovtsov2 Carsten Rother1 Bogdan Savchynskyy1 1 TU Dresden, Dresden, Germany 2 TU Graz, Graz, Austria |
| Pseudocode | No | The paper describes algorithms in prose but does not include any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper mentions re-implementing baselines and using existing solvers, but does not state that its own source code is being released or provide a link to it. |
| Open Datasets | Yes | We base our experiments on two datasets: (i) The interactive foreground/background image segmentation dataset utilized in several papers [4, 31, 22, 23] for comparing diversity techniques; (ii) A new dataset for foreground/background image segmentation with binary pairwise energies derived from the well-known PASCAL VOC 2012 dataset [11]. ... The authors of [4] kindly provided us their 50 super-pixel graphical model instances. They are based on a subset of the PASCAL VOC 2010 [11] segmentation challenge with manually added scribbles. ... The Pascal VOC 2012 [11] segmentation dataset has 21 labels. |
| Dataset Splits | No | The paper mentions "Parameters λ (from (7) and (2)) were tuned via cross-validation for each algorithm and each experiment separately." However, it does not provide specific details on how the dataset was split for training, validation, or testing, such as percentages or sample counts. |
| Hardware Specification | No | For the experiments we use a computer with 6 physical cores (12 virtual cores), and run Parametric-parallel with M threads. This is a general description of cores but lacks specific hardware details like CPU model, GPU model, or memory. |
| Software Dependencies | No | We utilize the dynamic graph-cut [24] technique for Parametric-sequential... The max-flow solver of [6] is used for Parametric-parallel together with Open MP directives. The paper mentions software components but does not provide specific version numbers for them. |
| Experiment Setup | Yes | Parameters λ (from (7) and (2)) were tuned via cross-validation for each algorithm and each experiment separately. |