# UniStitch: Unifying Semantic and Geometric Features for Image Stitching

arXiv: https://arxiv.org/abs/2603.10568 · HTML: https://mlsanity.com/paper/2603.10568/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 26 of its 53 references that are in the map.
- First posted on arXiv 11 Mar 2026.

## Builds on (most cited first, 26 of 26)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation](https://mlsanity.com/paper/1612.00593/) (2016)
- [PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space](https://mlsanity.com/paper/1706.02413/) (2017)
- [SuperGlue: Learning Feature Matching with Graph Neural Networks](https://mlsanity.com/paper/1911.11763/) (2019)
- [SuperPoint: Self-Supervised Interest Point Detection and Description](https://mlsanity.com/paper/1712.07629/) (2017)
- [PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume](https://mlsanity.com/paper/1709.02371/) (2017)
- [LightGlue: Local Feature Matching at Light Speed](https://mlsanity.com/paper/2306.13643/) (2023)
- [Point Transformer V3: Simpler, Faster, Stronger](https://mlsanity.com/paper/2312.10035/) (2023)
- [PointNeXt: Revisiting PointNet++ with Improved Training and Scaling Strategies](https://mlsanity.com/paper/2206.04670/) (2022)
- [YOLOv11: An Overview of the Key Architectural Enhancements](https://mlsanity.com/paper/2410.17725/) (2024)
- [Deep Image Homography Estimation](https://mlsanity.com/paper/1606.03798/) (2016)
- [Multi-Task Dense Prediction via Mixture of Low-Rank Experts](https://mlsanity.com/paper/2403.17749/) (2024)
- [Unsupervised Deep Image Stitching: Reconstructing Stitched Features to Images](https://mlsanity.com/paper/2106.12859/) (2021)
- [Parallax-Tolerant Unsupervised Deep Image Stitching](https://mlsanity.com/paper/2302.08207/) (2023)
- [Depth-Aware Multi-Grid Deep Homography Estimation with Contextual Correlation](https://mlsanity.com/paper/2107.02524/) (2021)
- [Single-Perspective Warps in Natural Image Stitching](https://mlsanity.com/paper/1802.04645/) (2018)
- [Thinking with Camera: A Unified Multimodal Model for Camera-Centric Understanding and Generation](https://mlsanity.com/paper/2510.08673/) (2025)
- [RecRecNet: Rectangling Rectified Wide-Angle Images by Thin-Plate Spline Model and DoF-based Curriculum Learning](https://mlsanity.com/paper/2301.01661/) (2023)
- [Object-level Geometric Structure Preserving for Natural Image Stitching](https://mlsanity.com/paper/2402.12677/) (2024)
- [MOWA: Multiple-in-One Image Warping Model](https://mlsanity.com/paper/2404.10716/) (2024)
- [Image Stitching Based on Planar Region Consensus](https://mlsanity.com/paper/2007.02722/) (2020)
- [Towards Robust Image Stitching: An Adaptive Resistance Learning against Compatible Attacks](https://mlsanity.com/paper/2402.15959/) (2024)
- [Learning Residual Elastic Warps for Image Stitching under Dirichlet Boundary Condition](https://mlsanity.com/paper/2309.01406/) (2023)
- [Parallax-tolerant Image Stitching via Segmentation-guided Multi-homography Warping](https://mlsanity.com/paper/2406.19922/) (2024)
- [StabStitch++: Unsupervised Online Video Stitching with Spatiotemporal Bidirectional Warps](https://mlsanity.com/paper/2505.05001/) (2025)
- [Robust Image Stitching with Optimal Plane](https://mlsanity.com/paper/2508.05903/) (2025)

## Abstract (arXiv)

Traditional image stitching methods estimate warps from hand-crafted geometric features, whereas recent learning-based solutions leverage semantic features from neural networks instead. These two lines of research have largely diverged along separate evolution, with virtually no meaningful convergence to date. In this paper, we take a pioneering step to bridge this gap by unifying semantic and geometric features with UniStitch, a unified image stitching framework from multimodal features. To align discrete geometric features (i.e., keypoint) with continuous semantic feature maps, we present a Neural Point Transformer (NPT) module, which transforms unordered, sparse 1D geometric keypoints into ordered, dense 2D semantic maps. Then, to integrate the advantages of both representations, an Adaptive Mixture of Experts (AMoE) module is designed to fuse geometric and semantic representations. It dynamically shifts focus toward more reliable features during the fusion process, allowing the model to handle complex scenes, especially when either modality might be compromised. The fused representation can be adopted into common deep stitching pipelines, delivering significant performance gains over any single feature. Experiments show that UniStitch outperforms existing state-of-the-art methods with a large margin, paving the way for a unified paradigm between traditional and learning-based image stitching.