Curriculum Learning with Diversity for Supervised Computer Vision Tasks

Abstract

Curriculum learning techniques are a viable solution for improving the accuracy of automatic models, by replacing the traditional random training with an easy to hard strategy. However, the standard curriculum methodology does not automatically provide improved results, but is constrained by multiple elements like the data distribution or the proposed model. In this paper, we introduce a novel curriculum sampling strategy which takes into consideration the diversity of the training data together with the difficulty of the inputs. We determine the difficulty using a state-of-the-art difficulty estimator and we model the diversity during training, giving higher priority to elements from classes visited less. We conduct object detection and instance segmentation experiments on Pascal VOC 2007 and Cityscapes data sets, surpassing both the randomly-trained baseline and the standard curriculum approach. We prove that our strategy is very efficient in unbalanced data sets, leading to faster convergence and more accurate results, where other curriculum-based strategies fail.