Hedgementation = Hedgerow Segmentation: A Remote Sensing Benchmark

Nathan Senyard, Salem Hamdani, Astrid Zhang, Derek Wang, Evan Shelhamer, Mathias Lécuyer, Joséphine Gantois, UBC CS · 2026

Read Hedgementation = Hedgerow Segmentation: A Remote Sensing Benchmark by Nathan Senyard, Salem Hamdani, Astrid Zhang, Derek Wang, Evan Shelhamer, Mathias Lécuyer, Joséphine Gantois, UBC CS (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-23615.txt.

Introduction

Senyard and colleagues release a benchmark for mapping hedgerows from satellite imagery across France, in a 2026 paper on remote sensing. Hedgerows, lines of shrubs and trees along field edges, support wildlife, store carbon, and protect soil. They have declined sharply, yet their extent is poorly mapped because field surveys are costly. Satellite images could fill the gap, but models must work in new regions and climates too.

The benchmark, called Hedgementation, merges harmonized Sentinel 2 imagery, learned image embeddings, and ground truth labels from a French national hedgerow inventory. Its splits test generalization across distance and across climatic zones, a harder challenge than random splits. Three baseline models are evaluated, covering supervised deep networks and a self supervised embedding approach. The best reaches above forty percent segmentation accuracy, stronger in farming areas and temperate zones.

Catalog record

CREATORNathan Senyard, Salem Hamdani, Astrid Zhang, Derek Wang, Evan Shelhamer, Mathias Lécuyer, Joséphine Gantois, UBC CS
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE52 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.23615

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Hedgementation = Hedgerow Segmentation: A Remote Sensing Benchmark Nathan Senyard, Salem Hamdani, Astrid Zhang, Derek Wang, Evan Shelhamer, Mathias Lécuyer, Joséphine Gantois, UBC CS Abstract We propose Hedgementation: a new benchmark to evaluate machine learning models for hedgerow mapping from remote sensing data at country scale and 10m 2 spatial resolution. We combine and harmonize multiple remote sensing data products and ground truth labels sourced from a hedgerow inventory in France. We measure the ability of three baseline models to generalize across spatial distance, and across climatic zones, a more explicitly challenging task. Our benchmark tests both supervised and self-supervised learning approaches for remote sensing, applied to tracking fine-scale features of high agricultural importance. The code to reproduce the benchmark and baselines results is available at https://github.com/hedgementation/hedgementation . 1 Introduction: Hedgerows and their Agricultural Importance Small patches of natural habitat like hedgerows play a critical role in agricultural landscapes, as they support a range of biodiversity outcomes, carbon storage, and ecosystem services ( 6 ; 2 ; 34 ; 1 ) . It is well-known that hedgerows have seen sharp and ongoing declines, in line with shifts in agricultural systems ( 5 ; 11 ) . Yet, little data exists on their remaining extent and dynamics, as it is costly to collect at large scales. Existing inventory efforts are geographically and temporally limited (two years of inventory in France and one year in England are the most prominent). Satellite imagery offers a promising solution to fill this data gap ( 20 ; 16 ) , especially when combined with deep learning models ( 31 ; 35 ) . However progress is limited by the sparsity of labels, unknown or poor transferability of existing models to new regions, and lack of access and reproducibility of the existing learning methods. We propose a new benchmark called Hedgementation, to evaluate hedgerow detection from remote sensing data across France (§ 2 ). Hedgementation builds a dataset of imagery, embeddings, and labels: it curates and combines harmonized imagery from the Sentinel-2 satellite ( 14 ) , embedding data from Alpha Earth Foundations ( 7 ) or AEF, and hedgerow labels from the BD Haie inventory produced by the French IGN institute ( 24 ) (§ 2.1 ). Based on this data, we design a benchmark for the hedgerow detection task. We partition the data to evaluate how models generalize across space and across agriculturally-relevant climatic zones (§ 2.2 ). We evaluate three baselines on Hedgementation (§ 3 ): two deep networks, U-TAE ( 18 ) and FTW 28 ), and one nearest neighbor model over the AEF embeddings. The initial results of the baselines show promise, with the best segmentation accuracy (intersection-over-union) reaching {>}40\% , and better performance in agricultural areas and in temperate zones, but underline the need for more progress to achieve accurate mapping. The code to automatically create our Hedgementation benchmark, and replicate our benchmark results, is publicly available at https://github.com/hedgementation/hedgementation . 2 Hedgementation Benchmark 2.1 The Hedgerow Dataset: Imagery, Embeddings, and Labels We combine several data sources to create a hedgerow dataset in France, where imagery and labels are publicly available. The satellite imagery comes from the harmonized, surface reflectance, Sentinel-2 data product ( 14 ) , accessed through Google Earth Engine (GEE) ( 19 ) . Sentinel-2 provides 10-m resolution images of multispectral bands at 5-day intervals. We use the 10 non-atmospheric bands, following prior work in self-supervised learning for remote sensing ( 17 ; 40 ) . We gather all images acquired Sept. 2021–Oct. 2022, and drop images with more than 20% cloud cover. The embeddings for Hedgementation are loaded from the AEF collection ( 7 ) on GEE. These are provided at a 10-m resolution in space, and at an annual resolution in time, where each “annualized” representation summarizes the full data over the calendar year. We collect the embeddings for 2022 for the highest overlap with the temporal interval of our satellite imagery. For any 10-m resolution pixel, we can thus load either the satellite imagery data product, with sensed spectral bands, or the embeddings data product, with learned bands. We pair these remote sensing inputs with hedgerow labels from version 2 of the French hedgerow inventory BD Haie ( 24 ) , which is unique in scope, accuracy, and public availability. These data are produced from aerial photo interpretation, validated by individual farmers in agricultural zones, and supplemented with automatic image segmentation in other areas ( 9 ) . They are representative of the status of hedgerows in 2020-2022 (Appendix A.1 ). The data cover all of France and contain 20,529,039 hedgerows, represented as line geometries with no width attribute. To match the 10-m resolution of the remote sensing inputs, we rasterize hedgerow lines through the following process, shown in Figure 1 : we apply a buffer around each line to create 7m-wide hedgerows 1 1 1 7 meters is the average width of hedgerows measured across Germany ( 13 ) , and is the width used in hedgerow accounting calculations by the French government ( 11 ) . ; rasterize buffered lines at a 0.5-m resolution (Figure 1(a) ); and downsample the resulting binary raster image to the target 10-m resolution by averaging (Figure 1(b) , 1(c) ). We split this national data into 1.28km x 1.28km square patches . This yields 603,564 data points (128x128 pixel images), which we eventually subsample to 2,995 for faster processing (§ 2.2 ). To capture differences in vegetation type and agricultural systems, we assign an agriculturally-relevant climatic zone to each patch. We use data on Global Agro-Ecological Zones ( 15 ) , available at a 10-km resolution, extract the thermal regime underlying the classification, and assign to each patch the regime that it overlaps the most with. We further group patches into a temperate zone (“Temperate, cool” and “Temperate, moderately cool” regimes) and subtropic zone (“Subtropics, cool” and “Subtropics, moderately cool” regimes), which comprise {\sim}99\% of the data. Finally, to evaluate whether the difficulty of the task varies inside and outside of agricultural zones, we use geospatial data on agricultural parcel boundaries across France ( 8 ) . 2.2 The Hedgerow Detection Task and its Evaluation The Hedgementation benchmark uses the hedgerow dataset to evaluate semantic segmentation approaches to hedgerow detection and assess model generalization across space and climatic zones. (a) Hedgerows rasterized at 0.5-m resolution. (b) Final label raster at 10-m resolution. (c) Final label raster overlaid on Sentinel-2 image. (d) Sentinel-2 image for the chosen patch. Figure 1: Hedgerow label rasterization process, applied to one patch : spatial detail is preserved. Assignment to train/val/test. We first group patches into cells of 12x12 patches separated by a 2-patch-wide buffer. We then randomly assign cells into one of five folds (see Figure 2(b) ), and partition the folds into three for training (60% of the data), one for validation (20% of the data) and one for testing (20% of the data). The spatial buffer between cells ensures that no validation patch is directly adjacent to a training patch, and no test patch is directly adjacent to a training or validation patch. This is important as hedgerows and other landscape features can cross patch boundaries. Evaluation metric. To evaluate

Read the complete text