Y Gao, Z Wang, WS Zheng, C Xie… - Proceedings of the …, 2024 - openaccess.thecvf.com
Contrastive learning has emerged as a promising paradigm for 3D open-world understanding ie aligning point cloud representation to image and text embedding space …
Mainstream 3D representation learning approaches are built upon contrastive or generative modeling pretext tasks, where great improvements in performance on various downstream …
Large foundation models have recently emerged as a prominent focus of interest, attaining superior performance in widespread scenarios. Due to the scarcity of 3D data, many efforts …
X Wu, X Wen, X Liu, H Zhao - Proceedings of the IEEE/CVF …, 2023 - openaccess.thecvf.com
As a pioneering work, PointContrast conducts unsupervised 3D representation learning via leveraging contrastive learning over raw RGB-D frames and proves its effectiveness on …
B Fei, Y Li, W Yang, L Ma, Y He - arXiv preprint arXiv:2404.13619, 2024 - arxiv.org
State-of-the-art 3D models, which excel in recognition tasks, typically depend on large-scale datasets and well-defined category sets. Recent advances in multi-modal pre-training have …
The success of deep learning heavily relies on large-scale data with comprehensive labels, which is more expensive and time-consuming to fetch in 3D compared to 2D images or …
H Li, Y Zhou, Y Zeng, H Xu, X Liang - arXiv preprint arXiv:2402.06198, 2024 - arxiv.org
3D Shape represented as point cloud has achieve advancements in multimodal pre-training to align image and language descriptions, which is curial to object identification …
The rising importance of 3D representation learning, pivotal in computer vision, autonomous driving, and robotics, is evident. However, a prevailing trend, which straightforwardly …
Pre-training by numerous image data has become de-facto for robust 2D representations. In contrast, due to the expensive data processing, a paucity of 3D datasets severely hinders …