Less is More: Towards Efficient Few-shot 3D Semantic Segmentation via Training-free Networks

TL;DR

This paper introduces TFS3D, a training-free approach for few-shot 3D semantic segmentation that reduces training time and domain gap issues, achieving comparable or better performance than previous methods.

Contribution

Proposes TFS3D, a training-free network for 3D segmentation, and TFS3D-T, a lightweight training-based variant with improved accuracy and efficiency.

Findings

TFS3D achieves comparable performance without training.

TFS3D-T improves state-of-the-art by +6.93% and +17.96% mIoU.

Training time is reduced by 90%.

Abstract

To reduce the reliance on large-scale datasets, recent works in 3D segmentation resort to few-shot learning. Current 3D few-shot semantic segmentation methods first pre-train the models on `seen' classes, and then evaluate their generalization performance on `unseen' classes. However, the prior pre-training stage not only introduces excessive time overhead, but also incurs a significant domain gap on `unseen' classes. To tackle these issues, we propose an efficient Training-free Few-shot 3D Segmentation netwrok, TFS3D, and a further training-based variant, TFS3D-T. Without any learnable parameters, TFS3D extracts dense representations by trigonometric positional encodings, and achieves comparable performance to previous training-based methods. Due to the elimination of pre-training, TFS3D can alleviate the domain gap issue and save a substantial amount of time. Building upon TFS3D, TFS3D-T only requires to train a lightweight query-support transferring attention (QUEST), which enhances the interaction between the few-shot query and support data. Experiments demonstrate TFS3D-T improves previous state-of-the-art methods by +6.93% and +17.96% mIoU respectively on S3DIS and ScanNet, while reducing the training time by -90%, indicating superior effectiveness and efficiency.