the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
BenthiCat: An opti-acoustic dataset for advancing benthic classification and habitat mapping
Abstract. Benthic habitat mapping is fundamental for understanding marine ecosystems, guiding conservation efforts, and supporting sustainable resource management. Yet, the scarcity of large, annotated datasets limits the development and benchmarking of machine learning models in this domain. This paper introduces a thorough multi-modal dataset, comprising about a million side-scan sonar (SSS) tiles collected along the coast of Catalonia (Spain), complemented by bathymetric maps and a set of co-registered optical images from targeted surveys using an autonomous underwater vehicle (AUV). Approximately 36000 of the SSS tiles have been manually annotated with segmentation masks to enable supervised fine-tuning of classification models. All the raw sensor data, together with mosaics, are also released to support further exploration and algorithm development. To address challenges in multi-sensor data fusion for AUVs, we spatially associate optical images with corresponding SSS tiles, facilitating self-supervised, cross-modal representation learning. Accompanying open-source preprocessing and annotation tools are provided to enhance accessibility and encourage research. This resource aims to establish a standardized benchmark for underwater habitat mapping, promoting advancements in autonomous seafloor classification and multi-sensor integration.
- Preprint
(52815 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on essd-2026-565', Anonymous Referee #1, 11 Aug 2026
-
RC2: 'Comment on essd-2026-565', Anonymous Referee #2, 08 Nov 2026
This article presents a multi-modal dataset including side-scan sonar and optical images. The paper is well written and mostly clear but it needs some clarifications/improvements on the presentation. One of the main issues is the name/abstract presenting a multi-modal dataset when there are 4 datasets (only one includes optical images) and the presentation throughout the paper of these 4 versions should be improved to be clear. Namely, every time a new dataset is introduced it should be immediately named, the summary on Section 6 is not enough. Another main issue is that the sensors used were changed throughout the surveys. This is not a problem per se but it's not clear if the repository includes this information and notes it in an easy way to understand what part of the data is collected with a sensor set and which part with the new sensor set (for the different modalities/changes).
Minor issues:
Line 112 is not clear if port and starboard would always have the same range (when varying ranges)
Line 162Ā in a couple of places it says no more than 5 m but then the survey was done between 3.5 and 4.5 m, please clarify
Table 4 should mention immediately why optical cameras had different orientations (or not mention it at all)
Line 195 it seems that calibration was done outside the water (but a few lines before it seems it was underwater), please clarify
Line 202 "During the survey, the orientation of the stereo pair was modified" - please clarify this was done once per day and not while each survey was being performed.Ā
Line 211 does the navigation include the Phins output or the result of the UKF?
Line 226 again it mentions 5 m and before 4.5 m
Line 244 how were these waterfalls chosen? Is it the good ones without many issues or the most representative?
Fig. 7 are these the manual masks or ML-based?
Between section 5.1 and 5.2, there might be the need of one section for MBES
Line 294 What about roll influence in the area covered?
Line 318 Final dataset should be named (as well as before for each version so it's clear).
Line 370 It would be better if the figures of Appendix A would go here before Appendix B
All Appendix A figures should refer the name of the corresponding dataset
Citation: https://doi.org/10.5194/essd-2026-565-RC2
Data sets
BenthiCat: An opti-acoustic dataset for advancing benthic classification and habitat mapping Hayat Rajani https://dataverse.harvard.edu/dataverse/benthicat
Model code and software
Opti-acoustic Pre-processing Tools Hayat Rajani and Valerio Franchi https://github.com/CIRS-Girona/opti-acoustic-preprocessing-tools
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 434 | 241 | 177 | 852 | 215 | 167 |
- HTML: 434
- PDF: 241
- XML: 177
- Total: 852
- BibTeX: 215
- EndNote: 167
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The manuscript entitled āBenthiCat: An opti-acoustic dataset for advancing benthic classification and habitat mappingā constitutes an excellent contribution to benthic habitat mapping and is highly recommended for publication. Ā The Objectives are clearly stated, and the Methodology is highly detailed.
My comments are very specific and presented below:
I believe the authors will be able to rapidly address these modifications and get the manuscript published in a short time.