Data for: Hydrophobicity Classification of Composite Insulators Based on Convolutional Neural Networks

Published: 28-03-2020| Version 1 | DOI: 10.17632/w2pg3k6c8m.1
Contributors:
v k,
Christos-Christodoulos Kokalis,
Thanos Tasakos,
Giorgos Siolas,
Ioannis Gonos

Description

By applying the spray method (IEC Standard 62073), about 4500 photos were collected and are available online, from all hydrophobicity classes using distilled water-ethyl alcohol as spraying solution. The pictures of the seven different hydrophobicity classes were split into three separate sets for each hydrophobicity class. The first one consisting of 400 instances of each class (400 × 7 = 2800 photos) was used for the training of the networks. The second one consisting of 100 instances of each class (100 × 7 = 700 photos) was used for the evaluation-validation of the learning course and the comparison of the different models. The last one with 122-165 different instances of each class (980 photos) was used for the final assessment of our chosen model.