Apr 14, 2023 · We design a 3D Swin transformer as our backbone network, which enables efficient self-attention on sparse voxels with linear memory complexity.

We present a pretrained 3D backbone, named Swin3D, that first-time outperforms all state-of-the-art methods on downstream 3D indoor scene understanding tasks.

A 3D Swin transformer is designed as the backbone network, which enables efficient self-attention on sparse voxels with linear memory complexity.

Apr 14, 2023 · Our method demonstrates the great potential of pretrained 3D backbones with fine-tuning for 3D understanding tasks. The code and models are ...

Sep 20, 2023 · We design a 3D Swin transformer as our backbone network, which enables efficient self-attention on sparse voxels with linear memory complexity.

In this paper, we present a pretrained 3D backbone, named Swin3D, which first outperforms all state-of-the-art methods in downstream 3D indoor scene ...

Apr 17, 2023 · Swin3D: A Pretrained Transformer Backbone for 3D Indoor Scene Understanding. AI · arxiv.org. Open. Upvote 12. Downvote 1 Go to comments

Swin3D: A Pretrained Transformer Backbone for 3D Indoor Scene Understanding · Updates · Introduction · Overview · Environment · Data Preparation · Training ...

Feb 22, 2024 · Swin3D is a novel 3D backbone architecture for point cloud analysis that adopts the Swin Transformer [23] model. It uses a hierarchical network ...

We design a 3D Swin transformer as our backbone network, which enables efficient self-attention on sparse voxels with linear memory complexity.