MANGO: A Global Single-Date Paired Dataset for Mangrove Segmentation

TL;DR

This work tackles the temporal pairing gap in mangrove monitoring by introducing MANGO, a public global dataset of 42,703 single-date Sentinel-2 image–mask pairs across 124 countries. The authors deploy a two-stage pipeline: scalable data collection via Google Earth Engine and adaptive MF-based single-date selection that maximizes separability between mangrove and background, producing high-quality image–mask pairs aligned with annual GMW labels. Benchmark results across seven segmentation backbones show that the MF-based selection yields superior IoU and F1 scores compared to an MVI-based baseline, promoting robust generalization under a country-disjoint split. By providing a reproducible, globally representative dataset and a rigorous evaluation protocol, MANGO enables scalable mangrove monitoring and more reliable climate-related assessments.

Abstract

Mangroves are critical for climate-change mitigation, requiring reliable monitoring for effective conservation. While deep learning has emerged as a powerful tool for mangrove detection, its progress is hindered by the limitations of existing datasets. In particular, many resources provide only annual map products without curated single-date image-mask pairs, limited to specific regions rather than global coverage, or remain inaccessible to the public. To address these challenges, we introduce MANGO, a large-scale global dataset comprising 42,703 labeled image-mask pairs across 124 countries. To construct this dataset, we retrieve all available Sentinel-2 imagery within the year 2020 for mangrove regions and select the best single-date observations that align with the mangrove annual mask. This selection is performed using a target detection-driven approach that leverages pixel-wise coordinate references to ensure adaptive and representative image-mask pairings. We also provide a benchmark across diverse semantic segmentation architectures under a country-disjoint split, establishing a foundation for scalable and reliable global mangrove monitoring.

Figures (5)

• Figure 1: Spectral-index instability. MVI responses for the same site at different dates ($t_1$, $t_2$) show significant drift despite visual similarities in RGB composites.
• Figure 2: Scene selection pipeline for constructing single-date image-mask pairs from annual mangrove labels. For each site, multiple Sentinel-2 candidates $\{I_{i,t}\}$ share the same annual mask $M_i$. We extract mangrove reference pixels from $M_i$ to form a target spectrum and compute a detection map $D_{i,t}$ using a background-whitened target detector. Each candidate is scored by the Fisher discriminant ratio $J(I_{i,t})$ over mangrove and background regions, and the final acquisition is selected by $t^*=\arg\max_t J(I_{i,t})$.
• Figure 3: Global overview and experimental setup of the MANGO dataset. (a) illustrates the categorization of tiles into positive and negative classes. (b) shows the distribution of MANGO images across diverse countries. (c) visualizes the geographic partition used for rigorous generalization testing.
• Figure 4: Qualitative results across baseline models. (a) Sentinel-2 L2A imagery, (b) MANGO ground-truth mask established via the quality-aware selection pipeline, and (c--i) segmentation predictions from baseline models.
• Figure 5: Selection comparison for two Sentinel-2 candidate observations ($t_1$ and $t_2$) from the same region. The mask is the shared GMW-derived annual label. MVI and MF maps show per-pixel responses, and the values in parentheses report the FDR score $J(I_{i,t})$ used to rank candidates.