Seabed images and programs for a Deep Learning training and testing on the MAR

Published: 29 December 2021| Version 1 | DOI: 10.17632/vxcv2dxscw.1
Contributor:
QK Zhao

Description

1. Datasets.zip This collection of images were part of that used as training and testing datasets for a a Deep Learning on the MAR in a paper that will published in Ocean Engineering recently. 2. Deep Learning programs.zip This file includes the program and related modules of the ResNet network model used. The file named train.py can be used for training, extract image features, and finally output weight information. Resnet34.pth is the pre-training weight. predict.py can be used to classify seabed images. 3. Weights.zip This file is the weight information output after the training, where Data_set_M.pth is the result of this example data set, and Data_set_L.pth is the final result of this experiment.

Files

Institutions

First Institute of Oceanography

Categories

Computer Vision, Image Registration, Mid-Ocean Ridge, Deep Learning, Hydrothermal Mineral Deposit, Ocean Mining

Licence