αDepth: Learning Single-Pass Soft Boundary Decomposition for Stereo Conversion
Summary
αDepth introduces a layered representation with Circular Alpha Representation (CAR) to address soft boundary challenges in stereo conversion, achieving state-of-the-art performance without manual guidance.
View Cached Full Text
Cached at: 06/03/26, 11:37 AM
Paper page - αDepth: Learning Single-Pass Soft Boundary Decomposition for Stereo Conversion
Source: https://huggingface.co/papers/2606.00386
Abstract
αDepth introduces a layered representation with Circular Alpha Representation (CAR) to address soft boundary challenges in stereo conversion through local boundary decomposition and efficient scene-level inference.
Accurately modelingsoft boundaries, e.g., hair and defocus blur, is a fundamental challenge instereo conversiondue to the ambiguous blending of foreground and background. Existing depth models primarily predict single-layer depth, leading to ambiguity in depth correspondence atsoft boundaries. Whilematting techniquescan capture opacity for layered modeling, they often struggle in complex scenes with multiple targets and usually require user intervention. This paper introduces αDepth, alayered representationthat decomposessoft boundariesfor high-fidelitystereo conversion. Specifically, we first resolve mixed color and depth ambiguity by estimating layered color and depth values atsoft boundaries. Considering complex multi-target scenes, we design aCircular Alpha Representation(CAR) that shifts the paradigm from global target extraction to local boundary decomposition. Unlike prior matting methods restricted to a single foreground/background, CAR enables efficientscene-level inferencewithout manual guidance. Extensive evaluations demonstrate that αDepth achieves state-of-the-art performance instereo conversion, eliminatingbackground bleedingandstructural distortionsatsoft boundaries.
View arXiv pageView PDFAdd to collection
Get this paper in your agent:
hf papers read 2606\.00386
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2606.00386 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2606.00386 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2606.00386 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
One Scene, Two Depths: Probing Geometric Ambiguity in Monocular Foundation Models
Introduces MultiDepth-3k, a benchmark to evaluate depth-layer preferences in monocular depth foundation models, and shows Laplacian Visual Prompting can alter reported depth layers, suggesting complementary geometric hypotheses exist across models.
chenxwh/depth-anything-v2
Depth Anything V2 is a monocular depth estimation model that significantly outperforms V1 in fine-grained details and robustness, offering faster inference and higher accuracy than SD-based models. It is available on Replicate under varying licenses.
@RuohanZhang76: Excited to introduce StereoPolicy, led by @EvansXuHan. StereoPolicy is an effective way to add geometric cues to modern…
Introduces StereoPolicy, a framework that leverages synchronized stereo image pairs to improve geometric reasoning for robot manipulation policies, avoiding the fragility of RGB-D and point clouds. It integrates with diffusion-based and vision-language-action policies, showing consistent improvements in simulation and real-world tasks.
Lite Any Stereo V2: Faster and Stronger Efficient Zero-Shot Stereo Matching
Lite Any Stereo V2 presents an efficient stereo matching approach achieving state-of-the-art accuracy with significantly reduced latency through optimized architecture and training strategies, including a 2D-only cost aggregation framework and a three-stage training strategy.
Patil/Krea-2-depth-controlnet
This model introduces a depth-conditioned ControlNet-LoRA for Krea-2, enabling depth-map-guided image generation with high depth consistency (0.98-0.99 Pearson correlation). It supports both Raw and Turbo variants and includes easy inference scripts and Comfy UI integration.