paper-with-me

홈 › Papers

Point Transformer V3: Simpler, Faster, Stronger

2023-12-15 · Xiaoyang Wu, Li Jiang, Peng-Shuai Wang, Zhijian Liu, Xihui Liu, Yu Qiao, Wanli Ouyang, Tong He, Hengshuang Zhao

This paper is not motivated to seek innovation within the attention mechanism. Instead, it focuses on overcoming the existing trade-offs between accuracy and efficiency within the context of point cloud processing, leveraging the power of scale. Drawing inspiration from recent advances in 3D large-scale representation learning, we recognize that model performance is more influenced by scale than by intricate design. Therefore, we present Point Transformer V3 (PTv3), which prioritizes simplicity and efficiency over the accuracy of certain mechanisms that are minor to the overall performance after scaling, such as replacing the precise neighbor search by KNN with an efficient serialized neighbor mapping of point clouds organized with specific patterns. This principle enables significant scaling, expanding the receptive field from 16 to 1024 points while remaining efficient (a 3x increase in processing speed and a 10x improvement in memory efficiency compared with its predecessor, PTv2). PTv3 attains state-of-the-art results on over 20 downstream tasks that span both indoor and outdoor scenarios. Further enhanced with multi-dataset joint training, PTv3 pushes these results to a higher level.

📄 PDF Abstract BibTeX arXiv:2312.10035

Code (3)

Pointcept/Pointcept 공식 구현 pytorch
pointcept/pointtransformerv3 공식 구현 pytorch
facebookresearch/SparseConvNet pytorch

Tasks

3D Semantic SegmentationLIDAR Semantic SegmentationRepresentation LearningSemantic Segmentation

Methods 이 논문이 사용한 방법론

Multi-Head Attention 설명 없음
Attention 설명 없음
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
Dense Connections Dense Connections, or Fully Connected Connections, are a type of layer in a deep neural network that use a linear operation where every input is connected to every output…
Position-Wise Feed-Forward Layer 설명 없음
Absolute Position Encodings Absolute Position Encodings are a type of position embeddings for [Transformer-based models] where positional encodings are…
Label Smoothing Label Smoothing is a regularization technique that introduces noise for the labels. This accounts for the fact that datasets may have mistakes in them, so maximizing the…
SPEED The monocular depth estimation (MDE) is the task of estimating depth from a single frame. This information is an essential knowledge in many computer vision tasks such as scene…

Similar Papers 제목 키워드 기반

PointOBB-v2: Towards Simpler, Faster, and Stronger Single Point Supervised Oriented Object Detection

2024-10-10 · Botao Ren, Xue Yang, Yi Yu, Junwei Luo 외

Single point supervised oriented object detection has gained attention and made initial progress within the community. Diverse from those approaches relying on one-shot samples or powerful pretrained models (e.g. SAM), P…

object-detectionObject DetectionOriented Object Detection

ResT V2: Simpler, Faster and Stronger

2022-04-15 · Qing-Long Zhang, Yu-Bin Yang

This paper proposes ResTv2, a simpler, faster, and stronger multi-scale vision Transformer for visual recognition. ResTv2 simplifies the EMSA structure in ResTv1 (i.e., eliminating the multi-head interaction part) and em…

Semantic Segmentation

Point Transformer V3: Simpler Faster Stronger

2024-01-01 · CVPR 2024 1 · Xiaoyang Wu, Li Jiang, Peng-Shuai Wang, Zhijian Liu 외

This paper is not motivated to seek innovation within the attention mechanism. Instead it focuses on overcoming the existing trade-offs between accuracy and efficiency within the context of point cloud processing lev…

Representation Learning

Efficient Linear Attention for Fast and Accurate Keypoint Matching

2022-04-16 · Suwichaya Suwanwimolkul, Satoshi Komorita

Recently Transformers have provided state-of-the-art performance in sparse matching, crucial to realize high-performance 3D vision applications. Yet, these Transformers lack efficiency due to the quadratic computational …

Image MatchingVisual Localization

LitePT: Lighter Yet Stronger Point Transformer

2025-12-15 · Yuanwen Yue, Damien Robert, Jianyuan Wang, Sunghwan Hong 외 arxiv

Modern neural architectures for 3D point cloud processing contain both convolutional layers and attention blocks, but the best way to assemble them remains unclear. We analyse the role of different computational blocks i…