Class-Level Confidence Based 3D Semi-Supervised Learning
This work addresses a common data imbalance issue in 3D semi-supervised learning, offering improvements for tasks like classification and detection, though it is incremental as it builds on prior methods like FlexMatch.
The paper tackles the problem of imbalanced data in 3D semi-supervised learning by proposing a class-level confidence method, which significantly outperforms state-of-the-art counterparts in classification and detection tasks across all datasets.
Recent state-of-the-art method FlexMatch firstly demonstrated that correctly estimating learning status is crucial for semi-supervised learning (SSL). However, the estimation method proposed by FlexMatch does not take into account imbalanced data, which is the common case for 3D semi-supervised learning. To address this problem, we practically demonstrate that unlabeled data class-level confidence can represent the learning status in the 3D imbalanced dataset. Based on this finding, we present a novel class-level confidence based 3D SSL method. Firstly, a dynamic thresholding strategy is proposed to utilize more unlabeled data, especially for low learning status classes. Then, a re-sampling strategy is designed to avoid biasing toward high learning status classes, which dynamically changes the sampling probability of each class. To show the effectiveness of our method in 3D SSL tasks, we conduct extensive experiments on 3D SSL classification and detection tasks. Our method significantly outperforms state-of-the-art counterparts for both 3D SSL classification and detection tasks in all datasets.