RockMapper

🚧UNDER CONSTRUCTION-GUI Forthcoming🚧

Interface for predicting and mapping benthic habitat (substrates) from any side-scan sonar mosaic.

Overview

RockMapper is an open-source Python interface for automatically predicting and mapping rock from any side-scan sonar mosaic(s). RockMapper leverages SegFormer pre-trained models fine-tuned with Segmentation Gym to automatically classify "rocky" substrate- defined as any hard particle greater than 2 mm in size. RockMapper uses the Udden-Wentworth grain size scale when classifying substrate into four classes: other (fines <2mm), gravel (2mm - 64mm), boulder/cobble (>64mm ), and bedrock (unfractured hard substrates) (Wentworth 1922).

Published Documentation

Journal Article

Wolfenkoehler, W. A., Bodine, C. S., & Long, J. M. (forthcoming). Evaluating RockMapper: A software to automatically map aquatic substrate from sonar images.

In this study RockMapper was evaluated by comparing output maps to those made by manual methods, allowing future substrate delineators to better understand the potential strengths or limitations of each approach. RockMapper mapping resulted in similar accuracy to manual methods, which ranged between 61-98%, while also saving ~ 11 minutes per hectare of manual delineation time. Additionally, RockMapper appeared to generalize well to new rivers, indicating reproducibility to unseen datasets.

Model Training Data

Wolfenkoehler, W., and Long, J., 2026, Substrate classifications of three rivers in eastern Oklahoma: U.S. Geological Survey data release, https://doi.org/10.5066/P1CZGXSF

Installation

1. Install Miniforge.
2. Open the Miniforge prompt.
3. Install PINGInstaller:

   pip install --force-reinstall pinginstaller
   

4. Install RockMapper.

   python -m pinginstaller rockmapper
   

Usage

1. Copy the following script to some location on your computer and name it "RockMapper.py":

'''
Copyright (c) 2025 Cameron S. Bodine
'''

#########
# Imports
import os, sys
import time, datetime

start_time = time.time()

# Set ROCKMAPPER utils dir
USER_DIR = os.path.expanduser('~')
GV_UTILS_DIR = os.path.join(USER_DIR, '.rockmapper')
if not os.path.exists(GV_UTILS_DIR):
    os.makedirs(GV_UTILS_DIR)

def gui():
    '''
    '''

    #################
    # NEED TO ADD GUI


    # FOR DEVELOPMENT
    #############################
    # Update Parameters
    seg_model = 'RockMapper_20251117_v2'
    inDir = r'Z:\scratch\202506_BrushyDeepKiamichi_Substrate\mosaics' #location of the sonar mosaic you wish to process
    mosaicFileType = '.tif'
    outDirTop = r'Z:\scratch' #output folder where you wish to place the RockMapper substrate shapefile
    projName = 'RockMapperTest'
    mapRast = False #do you want a raster output?
    mapShp = True #do you want a shapefile output? 

    epsg = 32615 # change to desired cooordinate system

    windowSize_m = (18, 18)
    window_stride = 6
    minArea_percent = 0.75
    threadCnt = 0.75

    predBatchSize = 30

    minPatchSize_m2 = 5 # Minimum patch size to keep in final shapefile, in square meters
    smoothShp = True # Smooth final shapefile polygons
    smoothTol_m = 0.3 # Smoothing tolerance in meters, higher = more smoothing

    deleteIntData = True



    ################
    # Run HabiMapper

    modelDir = os.path.join(GV_UTILS_DIR, 'models')

    # RockMapper
    if seg_model in ['RockMapper_20250628_v1', 'RockMapper_20251117_v2']:
        from rockmapper.rock_mapper import do_work

        modelDir = os.path.join(modelDir, seg_model)


        print('\n\nMapping habitat with ROCKMAPPER model...\n\n')
        do_work(
        inDir = inDir,
        outDirTop = outDirTop,
        projName = projName,
        mapRast = mapRast,
        mapShp = mapShp,
        epsg = epsg,
        windowSize_m = windowSize_m,
        window_stride = window_stride,
        minArea_percent = minArea_percent,
        threadCnt = threadCnt,
        mosaicFileType=mosaicFileType, 
        modelDir=modelDir,
        predBatchSize=predBatchSize,
        deleteIntData=deleteIntData,
        minPatchSize = minPatchSize_m2,
        smoothShp = smoothShp,
        smoothTol_m = smoothTol_m,
        )





    print("\n\nGrand Total Processing Time: ", datetime.timedelta(seconds = round(time.time() - start_time, ndigits=0)))
    return

if __name__ == "__main__":
    gui()

2. Open the file with Visual Studio Code.
3. Update the Parameters as necessary:

#############################
# Update Parameters
seg_model = 'RockMapper_20251117_v2' 
inDir = r'Z:\scratch\202506_BrushyDeepKiamichi_Substrate\mosaics' #location of the sonar mosaic you wish to process
mosaicFileType = '.tif'
outDirTop = r'Z:\scratch' #output folder where you wish to place the RockMapper substrate shapefile
projName = 'RockMapperTest'
mapRast = False #do you want a raster output?
mapShp = True #do you want a shapefile output? 

epsg = 32615 # change to desired cooordinate system

windowSize_m = (18, 18)
window_stride = 6
minArea_percent = 0.75
threadCnt = 0.75

predBatchSize = 30

minPatchSize_m2 = 5 # Minimum patch size to keep in final shapefile, in square meters
smoothShp = True # Smooth final shapefile polygons
smoothTol_m = 0.3 # Smoothing tolerance in meters, higher = more smoothing

deleteIntData = True

4. Open Mini Forge Prompt
5. type "conda activate rockmapper"
6. change the directory to the folder where the python script is located
7. Run the script by typing "python RockMapper.py"