Seminário do discente Mateus Silva

Seminário do discente Mateus Silva, dia 25/03/2021 as 14:00

Título: Leaf Shape Reconstruction and Damage Estimation using a U-Net-Based Conditional GAN

Resumo: The leaf damage estimation is a relevant research topic regarding both agricultural and ecological studies. Most approaches that seek for shape reconstruction work with parametric curves, lacking generality when treating leaves with different shapes. The techniques that use neural networks approach this issue as a regression problem, estimating the damage's numerical value without trying to estimate the original leaf form. In this work, we propose a procedure to predict the original leaf shape and calculate its defoliation based on a Conditional Generative Adversarial Network (Conditional GAN). We trained and validated the algorithm with a dataset with leaf images from 33 different species. Also, we tested the produced model in another dataset, containing images from leaves from 153 different species. Our results show that the proposed method overcame the currently applied strategies, both in precision and generalism.


PPGCC - Programa de Pós-Graduação em Ciência da Computação

Departamento de Computação  |  ICEB  |  Universidade Federal de Ouro Preto
Campus Universitário Morro do Cruzeiro  |  CEP 35400-000  |  Ouro Preto - MG, Brasil
Telefone: +55 31 3559-1641  |