2K Retrofit: Entropy-Guided Efficient Sparse Refinement for High-Resolution 3D Geometry Prediction

2026-03-20

AI Summary

Existing geometric foundation models struggle to efficiently process 2K high-resolution images due to excessive computational and memory demands, hindering their practical deployment. This work proposes a plug-and-play, general-purpose inference framework that enables efficient 2K inference for any geometric foundation model without requiring architectural modifications or retraining. The approach leverages a fast coarse prediction followed by an entropy-guided sparse refinement mechanism that selectively enhances only the regions with high uncertainty. For the first time, this method achieves both state-of-the-art accuracy and speed across multiple mainstream benchmarks, significantly narrowing the gap between high-resolution 3D vision research and real-world applications.

Technology Category

• Computer Vision: Other Foundations of Computer Vision
• Machine Learning: Deep Neural Architectures and Foundation Models
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

High-resolution geometric prediction is essential for robust perception in autonomous driving, robotics, and AR/MR, but current foundation models are fundamentally limited by their scalability to real-world, high-resolution scenarios. Direct inference on 2K images with these models incurs prohibitive computational and memory demands, making practical deployment challenging. To tackle the issue, we present 2K Retrofit, a novel framework that enables efficient 2K-resolution inference for any geometric foundation model, without modifying or retraining the backbone. Our approach leverages fast coarse predictions and an entropy-based sparse refinement to selectively enhance high-uncertainty regions, achieving precise and high-fidelity 2K outputs with minimal overhead. Extensive experiments on widely used benchmark demonstrate that 2K Retrofit consistently achieves state-of-the-art accuracy and speed, bridging the gap between research advances and scalable deployment in high-resolution 3D vision applications. Code will be released upon acceptance.

Problem

• high-resolution 3D geometry prediction
• computational scalability
• memory efficiency
• geometric foundation models
• 2K inference

Innovation

• 2K Retrofit
• entropy-guided refinement
• sparse refinement
• high-resolution 3D geometry
• foundation model scalability