Convolutional Networks for Segmentation and Detection of Agricultural Mushrooms

Loading...
Thumbnail Image

Date

2018-06-07

Authors

Olpin, Alexander J.

Journal Title

Journal ISSN

Volume Title

Publisher

University of Guelph

Abstract

Previous research into agricultural crop identification has used standard image processing techniques to locate crops within image data. In this work we conducted two sets of experiments to determine an appropriate method for segmenting and detecting agricultural mushrooms. In the first experiments we found that a standard Convolutional Neural Network was able to achieve an 87.5% average precision rating while a Fully Convolutional Network was able to achieve a 88.9% average precision rating in mushroom segmentation. In the second set of experiments, a Region-based Convolutional Neural Network was able to achieve a 92.1% average precision rating while a Region-based Fully Convolutional Network was able to achieve an 87.6% average precision rating in mushroom detection. The results of our work showed that a Fully Convolutional Network was able to infer more accurately for the purpose of mushroom segmentation. While a Region-based Convolutional Network more in the area of mushroom detection.

Description

Keywords

Machine Vision, Agriculture, Convolutional Neural Networks, Fully Convolutional Networks, Region-based Convolutional Networks

Citation