Curriculum Learning with Diversity for Supervised Computer Vision Tasks
This work addresses the challenge of improving curriculum learning for supervised computer vision tasks, particularly for unbalanced datasets, though it is incremental as it builds on existing curriculum methods.
The paper tackled the problem of curriculum learning not automatically improving model accuracy by introducing a novel curriculum sampling strategy that considers both input difficulty and training data diversity, achieving faster convergence and more accurate results on object detection and instance segmentation tasks with Pascal VOC 2007 and Cityscapes datasets.
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 it 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 estimator based on the human time required for solving a visual search task. We consider this kind of difficulty metric to be better suited for solving general problems, as it is not based on certain task-dependent elements, but more on the context of each image. We ensure the diversity during training, giving higher priority to elements from less visited classes. 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 for unbalanced data sets, leading to faster convergence and more accurate results, when other curriculum-based strategies fail.