Benchmark for Quantitative Evaluation of Assisted Object Segmentation Methods to Image Sequences

Evaluation of segmentation methods applied to image sequences consists in the analysis of such methods according to quantitative and/or qualitative criteria, usually driven to some application. Literature proposes several metrics for quantitative evaluation of object segmentation methods to image sequences, but it is still considered an open problem, since no one of the proposed metrics is considered the standard one. More, as the best of our knowledge, there is no method in literature that does computational quantitative evaluation of assisted methods to object segmentation in image sequence. This paper introduces a benchmark to do such quantitative evaluation. This evaluation is done according to several criteria such as the robustness of segmentation and the easiness to segment the objects through the sequence. Experimental results also evaluates the robustness of the watershed from propagated markers technique.