Rotten Fruit Detection Using a One Stage Object Detector

Pérez-Daniel, Karina Ruby; Fierro Radilla, Atoany Nazareth; Peñaloza Cobos, José Pablo

doi:10.1007/978-3-030-60887-3_29

Rotten Fruit Detection Using a One Stage Object Detector

Journal

Advances in Computational Intelligence

Lecture Notes in Computer Science

ISSN

0302-9743

1611-3349

Date Issued

2020

Author(s)

Pérez-Daniel, Karina Ruby

Facultad de Ingeniería - CampCM

Fierro Radilla, Atoany Nazareth

Facultad de Ingeniería - CampCM

Peñaloza Cobos, José Pablo

Facultad de Ingeniería - CampCM

Type

text::book::book part

DOI

10.1007/978-3-030-60887-3_29

URL

https://scripta.up.edu.mx/handle/20.500.12552/4086

Abstract

Digital images and computer sciences have become two powerful tools in several areas, such as astronomy, medicine, forensics, etc. In the last years, computer sciences are getting involved in agricultural and food science to decide based on estimated or actual parameters named features. Rottenness is the state of decomposing or decaying the quality of the fruit, which not only affects the taste and appearance but also modifies its nutritional composition, causing the presence of mycotoxins dangerous for humans. Nowadays, rottenness detection is carried out using human inspection or using Ultraviolet light to highlight spots of rottenness represented as fluorescence. Recent computer vision approaches address this problem using hyperspectral imaging systems. In this paper, we propose to use a one-stage object detector inspired by RetinaNet to detect whether a fruit is fresh or rotten. One of the main stages of RetinaNet is based on computing a multi-scale convolutional feature pyramid network on top of a backbone. Therefore, in this work, we analyze the performance of RetinaNet using different artificial neural networks as backbone to determine the highest accuracy for fruit and rottenness detection. The experiments were done using a dataset composed of 13599 images divided by 6 classes, 3 fresh fruits, and 3 rotten fruits. The performance evaluation considers the mean average precision in the detection and the inference time of tested backbone models. © 2020, Springer Nature Switzerland AG.