Wolfram Computation Meets Knowledge

Wolfram Summer School


Enrico Castro Grespan

Science and Technology

Class of 2018


Enrico Castro Grespan is a physics engineer at UASLP, and he’s about to finish his master’s in renewable energies at IER-UNAM.

Computational Essay: IR Thermography

Project: Rooftop Recognition for Solar Energy Potential

Project Overview

TextData[{“A neural net was trained and tested using a dataset which contained the aerial images of five different cities and their corresponding mask. A mask is a binary image where white pixels tell us where are the building rooftops and a black pixel for the rest of the image. A generator function to load a batch of files into the training was needed to handle the whole dataset. The neural net to trained was chosen from the Semantic Segmentation category at Wolfram Neural Net Repository. “, StyleBox[““Ademxapp Model A1 Trained on PASCAL VOC2012 and MS-COCO Data””, FontSlant -> “Italic”], ” net was modify for the specific needs of the dataset and the classes to return. A good correlation between the test images and their rooftops was obtained.”}]