Saved in:
Bibliographic Details
Main Authors: Zeng, Zhaojie, Wang, Yuesong, Luo, Yawei, Guan, Tao
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.20820
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916030949883904
author Zeng, Zhaojie
Wang, Yuesong
Luo, Yawei
Guan, Tao
author_facet Zeng, Zhaojie
Wang, Yuesong
Luo, Yawei
Guan, Tao
contents 2D Gaussian splatting provides an efficient explicit representation for image reconstruction, but existing methods still require costly per-image iterative optimization or rely on handcrafted priors for primitive allocation. We present AIR, a self-supervised feed-forward framework that amortizes iterative Gaussian fitting into a single network pass, eliminating per-image test-time optimization. AIR adopts a stage-wise residual architecture that progressively predicts additional Gaussian primitives from reconstruction residuals, together with an explicit Stage Control mechanism that activates new primitives only in under-reconstructed regions. A Predict--Optimize--Distill training strategy stabilizes multi-stage prediction by distilling short-horizon optimized Gaussian increments back into the predictor. The stabilized predictor is then jointly finetuned across stages and equipped with an image-adaptive quantizer for compact Gaussian storage. Experiments on Kodak and DIV2K show that AIR achieves better reconstruction quality than representative Gaussian-based baselines while reducing encoding time to 160--300\,ms. Code: https://github.com/whoiszzj/AIR.git
format Preprint
id arxiv_https___arxiv_org_abs_2605_20820
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle AIR: Amortized Image Reconstruction Framework for Self-Supervised Feed-Forward 2D Gaussian Splatting
Zeng, Zhaojie
Wang, Yuesong
Luo, Yawei
Guan, Tao
Computer Vision and Pattern Recognition
2D Gaussian splatting provides an efficient explicit representation for image reconstruction, but existing methods still require costly per-image iterative optimization or rely on handcrafted priors for primitive allocation. We present AIR, a self-supervised feed-forward framework that amortizes iterative Gaussian fitting into a single network pass, eliminating per-image test-time optimization. AIR adopts a stage-wise residual architecture that progressively predicts additional Gaussian primitives from reconstruction residuals, together with an explicit Stage Control mechanism that activates new primitives only in under-reconstructed regions. A Predict--Optimize--Distill training strategy stabilizes multi-stage prediction by distilling short-horizon optimized Gaussian increments back into the predictor. The stabilized predictor is then jointly finetuned across stages and equipped with an image-adaptive quantizer for compact Gaussian storage. Experiments on Kodak and DIV2K show that AIR achieves better reconstruction quality than representative Gaussian-based baselines while reducing encoding time to 160--300\,ms. Code: https://github.com/whoiszzj/AIR.git
title AIR: Amortized Image Reconstruction Framework for Self-Supervised Feed-Forward 2D Gaussian Splatting
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.20820