FILTR: Extracting Topological Features from Pretrained 3D Models

2026-04-24

AI Summary

This work addresses the challenge of efficiently extracting topological information from pretrained 3D point cloud encoders and generating persistence diagrams. The authors propose FILTR, a novel framework that formulates topological feature extraction as a set prediction task for the first time. Building upon frozen pretrained encoders—such as Point-BERT or Point-MAE—FILTR introduces a learnable Transformer decoder to end-to-end generate persistence diagrams directly from raw point cloud inputs. To facilitate evaluation, the authors construct DONUT, a synthetic benchmark with controllable topological complexity. Experiments demonstrate that, despite the limited global topological signals preserved by existing encoders, FILTR effectively leverages their features to approximate accurate persistence diagrams, thereby validating the feasibility and efficiency of the proposed approach.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: 3D Computer Vision
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Recent advances in pretraining 3D point cloud encoders (e.g., Point-BERT, Point-MAE) have produced powerful models, whose abilities are typically evaluated on geometric or semantic tasks. At the same time, topological descriptors have been shown to provide informative summaries of a shape's multiscale structure. In this paper we pose the question whether topological information can be derived from features produced by 3D encoders. To address this question, we first introduce DONUT, a synthetic benchmark with controlled topological complexity, and propose FILTR (Filtration Transformer), a learnable framework to predict persistence diagrams directly from frozen encoders. FILTR adapts a transformer decoder to treat diagram generation as a set prediction task. Our analysis on DONUT reveals that existing encoders retain only limited global topological signals, yet FILTR successfully leverages information produced by these encoders to approximate persistence diagrams. Our approach enables, for the first time, data-driven extraction of persistence diagrams from raw point clouds through an efficient learnable feed-forward mechanism.

Problem

• topological features
• 3D point clouds
• persistence diagrams
• pretrained encoders
• topological information

Innovation

• FILTR
• persistence diagrams
• topological features
• 3D point cloud encoders
• Transformer decoder