0% found this document useful (0 votes)
45 views11 pages

The Problem and Its Setting

This document introduces a study that aims to develop a computer vision system using convolutional neural networks to evaluate pork meat quality. It discusses that consumers' meat purchasing decisions are influenced more by appearance than other factors. The study seeks to automate meat quality evaluation by developing a device that can classify meat samples into categories of quality in an objective and non-destructive manner. The significance of the study is that it may help shorten meat inspection time and provide reliable quality information to consumers and administrators.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
45 views11 pages

The Problem and Its Setting

This document introduces a study that aims to develop a computer vision system using convolutional neural networks to evaluate pork meat quality. It discusses that consumers' meat purchasing decisions are influenced more by appearance than other factors. The study seeks to automate meat quality evaluation by developing a device that can classify meat samples into categories of quality in an objective and non-destructive manner. The significance of the study is that it may help shorten meat inspection time and provide reliable quality information to consumers and administrators.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 11

CHAPTER I

THE PROBLEM AND ITS SETTING

Introduction

Globally, pork is the most consumed meat (15.8 kg/capita/year), followed by poultry

(13.6 kg/capita/year), beef (9.6 kg/capita/year) and lastly sheep and goat meat (1.9

kg/capita/year) (Gokirmali and Bayram, 2017). Consumers’ meat purchasing decisions are

influenced more by product appearance, such as color and marbling, than any other quality

factor (Font-i-Furnols et al., 2015).

Presently, there are two methods used to evaluate meat quality. These are sensory

evaluation and objective evaluation. Sensory evaluation is carried out by highly trained

inspectors, involving the use of sight and smell. This can lead to inconsistent, subjective and

variable results due to differences in human response. Objective evaluation such as

chemical and microbiological analyses are more reliable than sensory evaluation. However,

they are destructive, tedious and time-consuming. This suggests that a new technology that

is rapid, accurate, non-destructive, and highly repeatable could be beneficial.

The potential of using computer vision system in the food industry has long been

recognized. Computer vision system tries to clone human assessment of color, content,

shape, and texture via its three main elements: camera, lighting, and image analysis

software. Hence, it has ability to inspect samples and analyze differences among samples in

an objective and non-destructive manner. Supported by powerful learning algorithms,

computer vision provides judgments accurately, quickly, and consistently. Learning


algorithms are employed to learn meaningful and nontrivial relationships automatically in a

set of training data and produce a generalization that can be used to classify data.

Among the applications of learning algorithms in computer vision for food quality

evaluation are classification and prediction of various food quality including pork (A. P. A. d.

C. Barbon et al., 2017) and (Sun, Young, Liu, & Newman, 2018), beef (Hosseinpour, Ilkhchi,

& Aghbashlo, 2019), and poultry (Ye, Iino, Zhang, & Oshita, 2015) and (Asmara, Rahutomo,

Hasanah, & Rahmad, 2017). Current research shows that computer vision is effective, rapid,

and objective technology for classification and prediction applicable to many food products.

With the help of learning algorithms, it has the potential to provide even better evaluations.

The aim of this study is to evaluate pork meat using computer vision system with

convolutional neural network and classify it into categories according to quality in an

objective and non-destructive manner.

Theoretical Framework

Research and development refers to the activities undertaken by individuals in

developing new services, products, or improving services or products. Research and

development constitutes the first stage of development of a potential new service or product.

According to Sugiyono (2014) research and development is a method used to produce

a certain product, and test the effectiveness of the product. The steps in this method are

usually referred as R & D cycle which consist of: (1) research analysis, needs assessment,

and proof of concept; (2) product planning and design; (3) preliminary product development;

(4) preliminary field testing; (5) product revision; (6) main field testing; (7) final product

revision; and (8) dissemination and implementation (Mufadhol, 2017). Most product
development projects can be divided into these stages and can be modelled using Input

Process Output.

The Input Process Output (IPO) model is a functional model of a general system. It

identifies the inputs (information, ideas, and resources used), processes (actions taken

using the inputs), and outputs (results of the processing).

This study also works on the theory of research and development. The research and

development cycle from Mufadhol (2017) are reflected in the Agile software development

methodology phases: (1) requirements definition, (2) planning, (3) designing, (4)

development, (5) release, and (6) track and monitor. After following these phases, a new

software product is developed and its effectiveness is evaluated using ISO25010 software

quality standards which has the criteria for functional suitability, performance efficiency,

usability, reliability, and portability. The construction of the device and the testing of its

effectiveness are clearly research and development activities as stated by Sugiyono (2014).
Research Paradigm

The research paradigm of this study found its basis from the above theoretical

framework and the concepts, theories and discussions in the Review of Related Literature.

Input Process Output


Knowledge requirements Agile Software
Meat Quality Evaluation Development
Computer Vision Methodology
Machine Learning Requirements Computer Vision System
Planning for Meat Quality
Software requirements Designing Evaluation with
Tensorflow Development Convolutional Neural
Release Network
Python Track and Monitor
Raspbian Buster Evaluation Criteria
Functional suitability
Hardware requirements: Performance Efficiency
Raspberry pi Usability
Picamera Reliability
LCD Portability

Feedback
Figure 1. Research Paradigm

Figure 1 shows the research paradigm of the study using the IPO model. The input

block consists of the knowledge requirements, software and hardware requirements.

Knowledge requirements are the basic information needed to understand in pursuing the

study. Software requirements involve knowledge and skills on certain computer

programming language and software library packages in order to create a workable system.
Hardware requirements include the materials and equipment to be used in the development

of the system.

The process block shows the Agile software development model stages undertaken to

create a working system.

The output block shows the developed computer vision system for meat quality

evaluation.

Statement of the Problem

The purpose of the study is to develop a device that will evaluate pork meat quality

using computer vision system with convolutional neural network.

Specifically, it seeks to answer the following research questions:

1. What are the stages undertaken in the development of the computer vision system for

meat quality evaluation using Agile methodology?

2. What is the evaluation of meat inspectors and IT practitioners of the computer vision

system for meat quality evaluation using ISO25010 in terms of:

2.1 functional suitability,

2.2 performance efficiency,

2.3 usability,

2.4 reliability, and

2.5 portability?
3. What is the difference on the evaluation of meat inspectors and IT practitioners of the

computer vision system for meat quality evaluation based on the aforementioned

criteria?

4. How reliable is the machine in terms of:

4.1 color,

4.2 texture,

4.3 exudation, and

4.4 pH?

5. What implications can be drawn based from the results of the study?

Hypothesis

There is no difference between the evaluations of meat inspectors from National Meat

Inspection Service (NMIS) and IT practitioners on the use of the computer vision system for

meat quality evaluation in terms of the aforementioned variables.

Significance of the Study

There were a variety of reasons why it was important to undertake this research.

These reasons are brought by the possibilities presented by technology use in pork meat

quality evaluation.

Meat inspectors play a vital role in keeping the meat in the market safe for

consumption and of quality. In this regard, providing an alternative tool that provides results

in a fast manner can at least shorten the time they spend on meat visual inspection

process.
The consumers can get an assurance that pork meat quality in the market are

checked using current imaging technology.

The NMIS administrators can translate the information derived from the results of this

study into effective planning and implementation of a new method of visual inspection of

pork meat.

Instrumentation engineers may compare existing objective measures with the

developed device’s results toward the creation of more standardized meat evaluation

protocols.

Future researchers of similar undertaking can find the study useful and may be used

as basis for their own design and development of a system for meat quality evaluation.

Scope and Delimitations of the Study

This study focused on the classification of meat according to quality based on color,

texture and exudation: PSE-pale, soft, exudative; RFN- reddish pink, firm, non-exudative;

and DFD-dark, firm, dry. The researcher developed a computer vision system for meat

quality evaluation which analyzed the quality of pork meat samples using convolutional

neural network architecture, specifically MobileNet. Open source scripts for retraining the

neural network was provided by Tensorflow. The device can perform objective, non-

destructive classification and can provide immediate results via a liquid crystal display

(LCD). These were implemented in the Raspberry pi platform running on Raspbian Buster

operating system.
The study was conducted in the meat sections of Silang Public Market, Tagaytay City

Market, and Kadiwa Market, all in the province of Cavite from August to November 2019.

The respondents of the study were meat inspectors from NMIS and IT practitioners.

Survey questionnaires were distributed after which data were collected, analyzed and

interpreted.
Definition of Terms

The following terms are defined operationally in order to have a clearer understanding in

this study.

Color refers to a phenomenon of light and visual perception that enables one to

differentiate otherwise identical objects.

Exudation refers to the degree of oozing of liquid substance from meat.

Functionality Suitability refers to the degree to which the device provides function

that meet stated and implied needs when used under specified conditions.

Performance Efficiency refers to the set of attributes that bear on the relationship

between the level of performance of the device and the amount of resources used, under

stated conditions.

pH refers to the degree of acidity of meat.

Portability refers to the degree of effectiveness and efficiency with which the device

can be transferred from one environment to the other.

Quality refers to the characteristics of meat that is acceptable to the consumers.

Reliability refers to the degree to which the device performs specified functions under

specified conditions for a specified period of time.

Usability refers to the degree the device can be used by specified users to achieved

specified goals with effectiveness, efficiency and satisfaction in a specified context of use.

Texture refers to the degree of firmness of meat.

You might also like