# BINO: Encoder Centric Self Supervised Stereo With Native Pair Input

Source: [BINO: Encoder Centric Self Supervised Stereo With Native Pair Input](<https://www.sota2.com/research/paper/69cb55340636e5bf4afca07c>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-01T05:00:12.613Z

Authors: Haokun Zhou

Published: 2026-03-29T23:26:09.000Z

## Abstract

Stereo needs features that preserve fine cross view correspondence rather than only semantic similarity. Recent self supervised vision models transfer well, but they are not built for this goal, and geometry focused methods often rely on a binocular decoder or another explicit linkage module during pretraining. BINO asks whether strong binocular structure can instead be learned inside a compact encoder. It does this by fusing the rectified pair at the input stage, forming stereo micro cell tokens, and using a row aware patch phase positional encoding. Training uses one view masked token only distillation together with occlusion and view specific appearance mismatch. In a strict low resource setting with pretraining only on KITTI object, BINO gives the best frozen descriptor results under a no linkage probe among all compared baselines on proxy dense stereo, hard negative retrieval, and KITTI Stereo~2012 disparity. With the same lightweight stereo head for every encoder, it stays near CroCo~v2 while using a much smaller encoder. Supplementary transfer experiments on KITTI Stereo~2015 show the same qualitative trend. These results suggest that much of the cross view reasoning often assigned to a separate linkage module can be learned inside a compact and reusable encoder.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Stereo Matching on KITTI Stereo 2012 (val)](<https://www.sota2.com/research/sota/stereo-matching-on-kitti-stereo-2012-val.md>): Stereo Matching; KITTI Stereo 2012 (val); EPE: 4.57.

- [Hard-negative Retrieval on KITTI object](<https://www.sota2.com/research/sota/hard-negative-retrieval-on-kitti-object.md>): Hard-negative retrieval; KITTI object; Top-1 Acc: 65.2.

- [Stereo Matching on KITTI object StereoSGBM reference (proxy)](<https://www.sota2.com/research/sota/stereo-matching-on-kitti-object-stereosgbm-reference-proxy.md>): Stereo Matching; KITTI object StereoSGBM reference (proxy); PCK@1tok: 84.07.

- [Stereo Matching on KITTI Stereo deterministic held-out 2015 (train)](<https://www.sota2.com/research/sota/stereo-matching-on-kitti-stereo-deterministic-held-out-2015-train.md>): Stereo Matching; KITTI Stereo deterministic held-out 2015 (train); Top-1 Accuracy: 70.

- [Stereo Matching on KITTI Stereo 2015 (train)](<https://www.sota2.com/research/sota/stereo-matching-on-kitti-stereo-2015-train.md>): Stereo Matching; KITTI stereo 2015 (train); GT WTA EPE: 4.742.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2603.27904>)

[PDF](<https://arxiv.org/pdf/2603.27904.pdf>)