Detecting trees near electric wires with Deep Learning

Published: 11 January 2021| Version 1 | DOI: 10.17632/zw4kstvxwp.1
Artur André Oliveira


Urban images and scripts used to train and test a deep leaning framework architecture to classify images as having or not intersections between trees and overhead powerlines/wires.


Steps to reproduce

The deep learning network scripts, weights, and other configuration files can be found at the `Deep Learning Folder`. In that folder, after installing the required packages (e.g. with `pip install -r requirements.txt') one can execute the `` script to train and test the provided network named `mobilenetv2_test39`. The `Test_Data` and `Training_Data` folders contain the images and label files (summan.smr) to be executed with the script.


Universidade de Sao Paulo Instituto de Matematica e Estatistica


Image Classification, Urban Analysis