Name: Daniel Buscombe
Type: User
Company: @MARDAScience mardascience.com
Bio: Human, geoscientist, and software developer @MardaScience. I make measurements from gridded data using ML. I develop in public, mostly in various GH orgs
Twitter: magic_walnut
Location: Flagstaff, AZ
Blog: https://www.mardascience.com/
Daniel Buscombe's Projects
Matlab GUI for semi-automation and annotation of deep-sea ROV benthic imagery
A repository to host the Journal of Open Source Software manuscript that describes the CoastSeg project
Geospatial data viewer which aims to visualize different types of data collected from USGS surveys (e.g. bathymetry, topography, grain size) and plot how the data changes over time. πΊοΈπποΈ
Buscombe & Ritchie (2018) Landscape Classification with Deep Neural Networks. Geosciences 2018, 8(7), 244
A python/openCV tool for "Human-In-The-Loop" image segmentation and label image creation using semi-supervised ML
Rapid image labeling for data-driven Earth science discovery.
Map of places I've conducted field (and lab) work to date. Click on mymap.geojson to see the interactive map :ocean:
Data and supplemental video for "Automated riverbed sediment classification using low-cost sidescan sonar." By Buscombe, Grams and Smith, J. Hydraulic Engineering 2015
Kivy/Python demonstration prototype video acquisition software for the ICES Innovation Fund 2015
Simple matlab toolbox for suspended sediment concentration estimates from acoustic backscatter profiler
A browser-based tool for Doodling on NAIP imagery to make morpho-sedimentary roughness maps
Zoo implementation for west coast usgs planecam imagery
web version of pyDGS for DGS-Online web application
python module for spatially explicit spectral analysis
grand canyon sandbar segmentation tool
model demonstration segments 15-m Landsat-7/8 or 10-m Sentinel-2 imagery into water/no water
Benchmarking of satellite-derived shoreline mapping techniques
Experimental implementation of unsupervised SDS using FeatUp
A neural gym for training deep learning models to carry out geoscientific image segmentation. Works best with labels generated using https://github.com/Doodleverse/dash_doodler
Semantic Segmentation Suite in TensorFlow. Modified for RS aerial imagery
streamlit app for binary segmentation
new website august 2016