![[2602.13588] Two-Stream Interactive Joint Learning of Scene Parsing and Geometric Vision Tasks](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.13588] Two-Stream Interactive Joint Learning of Scene Parsing and Geometric Vision Tasks
Summary
The paper presents TwInS, a novel framework for joint learning of scene parsing and geometric vision tasks, inspired by the human visual system's dual processing streams.
Why It Matters
This research addresses the challenges of scene understanding and geometric vision by proposing a unified approach that enhances performance without relying on expensive human-annotated data. It contributes to advancements in computer vision, potentially improving applications in robotics and AI.
Key Takeaways
- TwInS integrates scene parsing and geometric vision tasks through a dual-stream architecture.
- The framework utilizes a semi-supervised training strategy to leverage large-scale multi-view data.
- Extensive experiments demonstrate TwInS's superior performance over existing methods.
- The approach eliminates the need for costly human-annotated ground truth.
- Source code will be publicly available, promoting further research and application.
Computer Science > Computer Vision and Pattern Recognition arXiv:2602.13588 (cs) [Submitted on 14 Feb 2026] Title:Two-Stream Interactive Joint Learning of Scene Parsing and Geometric Vision Tasks Authors:Guanfeng Tang, Hongbo Zhao, Ziwei Long, Jiayao Li, Bohong Xiao, Wei Ye, Hanli Wang, Rui Fan View a PDF of the paper titled Two-Stream Interactive Joint Learning of Scene Parsing and Geometric Vision Tasks, by Guanfeng Tang and 7 other authors View PDF HTML (experimental) Abstract:Inspired by the human visual system, which operates on two parallel yet interactive streams for contextual and spatial understanding, this article presents Two Interactive Streams (TwInS), a novel bio-inspired joint learning framework capable of simultaneously performing scene parsing and geometric vision tasks. TwInS adopts a unified, general-purpose architecture in which multi-level contextual features from the scene parsing stream are infused into the geometric vision stream to guide its iterative refinement. In the reverse direction, decoded geometric features are projected into the contextual feature space for selective heterogeneous feature fusion via a novel cross-task adapter, which leverages rich cross-view geometric cues to enhance scene parsing. To eliminate the dependence on costly human-annotated correspondence ground truth, TwInS is further equipped with a tailored semi-supervised training strategy, which unleashes the potential of large-scale multi-view data and enables continuous s...