# Tiling and Stitching Segmentation Output for Remote Sensing: Basic Challenges and Recommendations

arXiv: https://arxiv.org/abs/1805.12219 · HTML: https://mlsanity.com/paper/1805.12219/

## Key facts

- Cited by 24 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 30 May 2018.

## Built on by (most cited first, 24 of 24)

- [GEO-Bench: Toward Foundation Models for Earth Monitoring](https://mlsanity.com/paper/2306.03831/) (2023)
- [Mission Critical -- Satellite Data is a Distinct Modality in Machine Learning](https://mlsanity.com/paper/2402.01444/) (2024)
- [Low-field magnetic resonance image enhancement via stochastic image quality transfer](https://mlsanity.com/paper/2304.13385/) (2023)
- [OAM-TCD: A globally diverse dataset of high-resolution tree cover maps](https://mlsanity.com/paper/2407.11743/) (2024)
- [A 3D Conditional Diffusion Model for Image Quality Transfer -- An Application to Low-Field MRI](https://mlsanity.com/paper/2311.06631/) (2023)
- [How Shift Equivariance Impacts Metric Learning for Instance Segmentation](https://mlsanity.com/paper/2101.05846/) (2021)
- [Mapping industrial poultry operations at scale with deep learning and aerial imagery](https://mlsanity.com/paper/2112.10988/) (2021)
- [Unsupervised Flood Detection on SAR Time Series](https://mlsanity.com/paper/2212.03675/) (2022)
- [Deploying Geospatial Foundation Models in the Real World: Lessons from WorldCereal](https://mlsanity.com/paper/2508.00858/) (2025)
- [PRUE: A Practical Recipe for Field Boundary Segmentation at Scale](https://mlsanity.com/paper/2603.27101/) (2026)
- [Tiling artifacts and trade-offs of feature normalization in the segmentation of large biological images](https://mlsanity.com/paper/2503.19545/) (2025)
- [COSMIC: Compress Satellite Images Efficiently via Diffusion Compensation](https://mlsanity.com/paper/2410.01698/) (2024)
- [Improving Fabrication Fidelity of Integrated Nanophotonic Devices Using Deep Learning](https://mlsanity.com/paper/2303.12136/) (2023)
- [GAN-based synthetic FDG PET images from T1 brain MRI can serve to improve performance of deep unsupervised anomaly detection models](https://mlsanity.com/paper/2505.07364/) (2025)
- [Mapping Farmed Landscapes from Remote Sensing](https://mlsanity.com/paper/2506.13993/) (2025)
- [Remote SAMsing: From Segment Anything to Segment Everything](https://mlsanity.com/paper/2605.00256/) (2026)
- [From Machine Learning to Large-Scale EO Products: Best Practices for Making Maps](https://mlsanity.com/paper/2607.24532/) (2026)
- [Localise to segment: crop to improve organ at risk segmentation accuracy](https://mlsanity.com/paper/2304.04606/) (2023)
- [Streamlined Global and Local Features Combinator (SGLC) for High Resolution Image Dehazing](https://mlsanity.com/paper/2304.13375/) (2023)
- [Monitoring of Urban Changes with multi-modal Sentinel 1 and 2 Data in Mariupol, Ukraine, in 2022/23](https://mlsanity.com/paper/2309.08607/) (2023)
- [A Concise Tiling Strategy for Preserving Spatial Context in Earth Observation Imagery](https://mlsanity.com/paper/2404.10927/) (2024)
- [Per-channel autoregressive linear prediction padding in tiled CNN processing of 2D spatial data](https://mlsanity.com/paper/2502.12300/) (2025)
- [Semantic segmentation of forest stands using deep learning](https://mlsanity.com/paper/2504.02471/) (2025)
- [A Deep U-Net Framework for Flood Hazard Mapping Using Hydraulic Simulations of the Wupper Catchment](https://mlsanity.com/paper/2604.21028/) (2026)

## Abstract (arXiv)

In this work we consider the application of convolutional neural networks (CNNs) for pixel-wise labeling (a.k.a., semantic segmentation) of remote sensing imagery (e.g., aerial color or hyperspectral imagery). Remote sensing imagery is usually stored in the form of very large images, referred to as "tiles", which are too large to be segmented directly using most CNNs and their associated hardware. As a result, during label inference, smaller sub-images, called "patches", are processed individually and then "stitched" (concatenated) back together to create a tile-sized label map. This approach suffers from computational ineffiency and can result in discontinuities at output boundaries. We propose a simple alternative approach in which the input size of the CNN is dramatically increased only during label inference. This does not avoid stitching altogether, but substantially mitigates its limitations. We evaluate the performance of the proposed approach against a vonventional stitching approach using two popular segmentation CNN models and two large-scale remote sensing imagery datasets. The results suggest that the proposed approach substantially reduces label inference time, while also yielding modest overall label accuracy increases. This approach contributed to our wining entry (overall performance) in the INRIA building labeling competition.