English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

The rapid advancement of deep learning models often attributes to theirability to leverage massive training data. In contrast, such privilege has notyet fully benefited 3D deep learning, mainly due to the limited availability oflarge-scale 3D datasets. Merging multiple available data sources and lettingthem collaboratively train a single model is a potential solution. However, dueto the large domain gap between 3D point cloud datasets, such mixed supervisioncould adversely affect the model's performance and lead to degeneratedperformance (i.e., negative transfer) compared to single-dataset training. Inview of this challenge, we introduce Point Prompt Training (PPT), a novelframework for multi-dataset synergistic learning in the context of 3Drepresentation learning that supports multiple pre-training paradigms. Based onthis framework, we propose Prompt-driven Normalization, which adapts the modelto different datasets with domain-specific prompts and Language-guidedCategorical Alignment that decently unifies the multiple-dataset label spacesby leveraging the relationship between label text. Extensive experiments verifythat PPT can overcome the negative transfer associated with synergisticlearning and produce generalizable representations. Notably, it achievesstate-of-the-art performance on each dataset using a single weight-shared modelwith supervised multi-dataset training. Moreover, when served as a pre-trainingframework, it outperforms other pre-training approaches regardingrepresentation quality and attains remarkable state-of-the-art performanceacross over ten diverse downstream tasks spanning both indoor and outdoor 3Dscenarios.

Towards Large-scale 3D Representation Learning with Multi-dataset Point Prompt Training