Surveying the Deep: A Review of Computer Vision in the Benthos

This repository hosts the code used to reproduce figures found in the paper Surveying the Deep: A Review of Computer Vision in the Benthos (Trotter et al. 2025). For this paper, two figures (Figure 1 and Figure 7) were generated programmatically. The code is provided here for reproducibility purposes.

Instructions

To reproduce the aformentioned figures, please follow these steps.

1. Download the required CSV files

Download both the paper techniques (link) and literature latitude and longitude (link) CSV files from the paper's supplementary material. The former CSV is used to generate Figure 1, and the latter is used to generate Figure 7.

2. Generate a Python environment

Generate a Python environment using requirements.yaml. This was generated using mamba, so it is recommended you also use this to create the environment. If you don't want to use a virtual environment, provided you have geopandas==0.14.4 and its dependencies installed, the code should work. The code has been tested with Python 3.13.1.

mamba env create --file requirements.yaml
mamba activate surveying_the_deep

3. Run the code

To reproduce Figure 1, showing the progression of computer vision-based benthic biodiversity monitoring literature over time, run:

python3 techniques.py /path/to/techniques.csv $output_path                         

To reproduce Figure 7, showing the geographic origin of image data used to train the reviewed automated benthic image analysis systems, run:

python3 heatmap.py /path/to/lat_longs.csv $output_path

Running both scripts without optional arguments will produce the figures as they appear in the paper. Optional arguments are provided for customisation, please see the help message of each script for more information.

python3 techniques.py --help
python3 heatmap.py --help

Citation

@article{trotterSurveyingDeepReview2025,
  title = {Surveying the Deep: {{A}} Review of Computer Vision in the Benthos},
  shorttitle = {Surveying the Deep},
  author = {Trotter, Cameron and Griffiths, Huw J. and Whittle, Rowan J.},
  year = {2025},
  journal = {Ecological Informatics},
  pages = {102989},
  issn = {15749541},
  doi = {10.1016/j.ecoinf.2024.102989}