The GIST and RIST of Iterative Self-Training for Semi-Supervised Segmentation
This addresses a specific issue in semi-supervised segmentation for computer vision applications, but it is incremental as it builds on existing self-training methods.
The paper tackles the problem of performance degradation in iterative self-training for semi-supervised semantic segmentation by proposing GIST and RIST strategies, which alternate between human-labeled and pseudo-labeled data, resulting in a performance boost.
We consider the task of semi-supervised semantic segmentation, where we aim to produce pixel-wise semantic object masks given only a small number of human-labeled training examples. We focus on iterative self-training methods in which we explore the behavior of self-training over multiple refinement stages. We show that iterative self-training leads to performance degradation if done naïvely with a fixed ratio of human-labeled to pseudo-labeled training examples. We propose Greedy Iterative Self-Training (GIST) and Random Iterative Self-Training (RIST) strategies that alternate between training on either human-labeled data or pseudo-labeled data at each refinement stage, resulting in a performance boost rather than degradation. We further show that GIST and RIST can be combined with existing semi-supervised learning methods to boost performance.