BA-T: An Iterative Transformer for Two-View Bundle Adjustment

Hugging Face Daily Papers Papers

Summary

BA-T is an iterative Transformer architecture for two-view bundle adjustment that improves 3D reconstruction accuracy and cross-view consistency using a lightweight design with only 16% of conventional decoder parameters, matching or surpassing larger models.

Feed-forward models for 3D reconstruction have achieved strong performance using deep cross-view attention to exchange information across images. However, these approaches often depend on heavy decoder stacks and lack a structured mechanism for geometry refinement, resulting in poor multi-view consistency. We address this by drawing inspiration from classical bundle adjustment (BA), which can be viewed as an iterative information propagation process between poses and local geometry. Inspired by BA, we propose BA-T, an iterative Transformer that implements BA-style structured updates as a repeatable layer in implicit token space. Instead of relying on deep attention stacks, BA-T refines predictions based on latent residual by a single lightweight layer. Experiments demonstrate that BA-T progressively improves pose and reconstruction accuracy across iterations, achieves stronger cross-view consistency than conventional decoders, and matches or surpasses substantially larger models while using only 16% of their decoder parameters. BA-T provides a compact, efficient, and structural alternative to depth-heavy attention, enabling accurate 3D reconstruction within a lightweight architecture. The code will be made publicly at https://github.com/zhangganlin/BA-T.
Original Article
View Cached Full Text

Cached at: 06/03/26, 03:38 PM

Paper page - BA-T: An Iterative Transformer for Two-View Bundle Adjustment

Source: https://huggingface.co/papers/2606.03287

Abstract

BA-T is an iterative Transformer architecture that improves 3D reconstruction accuracy and cross-view consistency through structured updates inspired by bundle adjustment, using a lightweight design that requires only 16% of conventional decoder parameters.

Feed-forward modelsfor3D reconstructionhave achieved strong performance usingdeep cross-view attentionto exchange information across images. However, these approaches often depend on heavydecoder stacksand lack a structured mechanism forgeometry refinement, resulting in poor multi-view consistency. We address this by drawing inspiration from classicalbundle adjustment(BA), which can be viewed as an iterative information propagation process between poses and local geometry. Inspired by BA, we propose BA-T, aniterative Transformerthat implements BA-style structured updates as a repeatable layer inimplicit token space. Instead of relying on deep attention stacks, BA-T refines predictions based onlatent residualby a single lightweight layer. Experiments demonstrate that BA-T progressively improves pose and reconstruction accuracy across iterations, achieves strongercross-view consistencythan conventional decoders, and matches or surpasses substantially larger models while using only 16% of their decoder parameters. BA-T provides a compact, efficient, and structural alternative todepth-heavy attention, enabling accurate3D reconstructionwithin a lightweight architecture. The code will be made publicly at https://github.com/zhangganlin/BA-T.

View arXiv pageView PDFGitHub5Add to collection

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2606.03287 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2606.03287 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2606.03287 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Lite3R: A Model-Agnostic Framework for Efficient Feed-Forward 3D Reconstruction

Hugging Face Daily Papers

Lite3R is a model-agnostic framework that improves the efficiency of transformer-based 3D reconstruction using sparse linear attention and FP8-aware quantization. It reduces latency and memory usage by up to 2.4x while maintaining geometric accuracy on backbones like VGGT and DA3-Large.

Linearizing Vision Transformer with Test-Time Training

Hugging Face Daily Papers

This paper proposes a method to convert pretrained Softmax attention models into linear-complexity Test-Time Training (TTT) architectures, achieving comparable text-to-image quality to fine-tuned Softmax models while significantly accelerating inference. The approach is validated by linearizing Stable Diffusion 3.5, resulting in SD3.5-T^5 with 1.32x speedup at 1K resolution.