Skip to main content

Table 1 Mean pixel level accuracies for various algorithms on MSRC-21.

From: Scene Segmentation with Low-Dimensional Semantic Representations and Conditional Random Fields

Algorithm

LRC/NP*

LRC/LRC/NP*

LRC/LRC/MRF*

LRC/CRF/NP*

LRC/CRF/OS*

 

Accuracy (%)

72.7

75.4

76.7

75.9

76.8

 

Algorithm

TextonBoost [7]

PLSA-MRF [13]

Mean shift [9]

STF-ILP [19]

AC (ACP) [18]

RLP-CRF [21]*

Accuracy (%)

72.2

73.5

75.1

72

74.5 (77.7)

76.5

  1. *These methods give results averaged over five random train/test partitions. The others give results for a single partition.