Unsupervised quality control of segmentations based on a smoothness and intensity probabilistic model

•Unsupervised segmentation quality control to detect challenging segmentations produced by a set of raters.•Method based on the comparison between ground truth segmentations and segmentations produced by a smoothness and intensity probabilistic model.•Definition of a set of spatial priors for image segmentation and their Bayesian inference to estimate their parameters.•Estimation of the algorithm performances on 4 databases of medical and computer vision images and comparison with state of the art indices.