Agricultural Research Management Information System

Home
Research Summary
Search
About Us
- ARMIS
- Brochure
Contact Us
Report
User Request
Data Input
Help
- Operation Manual
  - PDF
  - Video
- Program Area & Commodity
We have reached 37600 number of research entries at this moment.

- Logout

Research Detail

Home
Research
Detail

Computer Vision Based Local Fruit Recognition

Md. Robel Mia
Department of CSE, Daffodil International University, Dhaka, Bangladesh

Md. Jueal Mia
Department of CSE, Daffodil International University, Dhaka, Bangladesh

Anup Majumder
Department of CSE, Jahangirnagar University, Dhaka, Bangladesh

Soummo Supriya
Department of CSE, Daffodil International University, Dhaka, Bangladesh

Md. Tarek Habib
Department of CSE, Jahangirnagar University, Daffodil International University, Bangladesh

Abstract/ Executive Summary:

Bangladesh is an agricultural country having a tropical monsoon climate. A large variety of tropical and sub-tropical fruits abound in Bangladesh. People of Bangladesh are fruit-lovers too. Currently, most of the people of this country are failing to recognize many of the rare local fruits and the number of this portion of people is increasing day by day. Thus, not only the natural heritage but also good sources of food are being diminished. Performing a machine vision based recognition of these fruits can help people recognize them. In this paper, we perform an in-depth exploration of a computer vision approach for recognizing rare local fruits of Bangladesh. A number of rare local fruits are classified based on the features extracted from their images. For our experiment, we have used a total of 480 images of 6 rare local fruits. We perform some preprocessing on the captured image and then expected features are extracted using image segmentation. Classification of the fruits is accomplished using support vector machines (SVMs). We have achieved 94.79% classification accuracy, which is not only good but also promising for future research.

Keywords: Computer vision, Feature extraction, Image segmentation, Local fruit, Performance metrics, Support vector machine (SVM)

Location: In Bangladesh

Start Date:

End Date:

Program Area: Knowledge Management

Commodity: Fruit, Remote sensing

Objectives:

In this paper, a computer vision based expert system is proposed which works on the captured image through mobile or handheld device and determines what fruits it is. Among a lot of local fruits, we have chosen 6 fruits, namely amla (amloki), sugar-apple (ata), bilombo, elephant apple (chalta), orboroi, sapota (sopheda) for our experiment. We perform preprocessing on the captured image than expected feature is extracted using segmentation. After segmentation, classification of fruits is done using support vector machines (SVMs).

Methodology:

The system architecture of a computer vision-based expert system for recognizing the rare fruits. It arises with the assumption that someone, i.e. a person who does not know the names of the fruit properly or a user can capture an image of fruit using a smartphone or device, where proposed system application is already installed. Then using mobile application user will send the captured image to our proposed expert system which is placed in the back-end server. From the front-end software or application, image is received through the internet. Then it is sent to the back-end server, where our expert system is installed. Based on the input image, the system sends its acknowledgment to the user. The user can find the feedback from the front-end mobile application.

There are numerous steps in image processing which are being used for classification of fruits. A machine vision based solution framework for the recognition of fruits. Our expert system starts with the image of fruit. The image of the fruit is obtained locally and from the Internet. Firstly, using bicubic interpolation the color image of fruit is transformed into a fixed-sized image. Let us assume that I is the intensity.

The accomplished feature vectors are applied to SVMs. Classification of data is a frequent task in machine learning. SVMs is a machine learning algorithm. It is also called linear classifier. It plays an important role in the field of biological science. SVMs is established to provide huge accuracy and ordinarily betters than other classifiers. SVMs basically designed for the classification of binary problems, but its handle multiclass problem too that is our main problem. If the classifiers produce binary output, two basic strategies are used. In the first strategy, classifier among one and the K − 1 other classes are trained (in entire classifiers K). This is called the 1-r (one-against-rest) approach. The one-against-rest (1-r) approaches including the number of SVMs can be used to overcome this type of problem. By using training data set, SVMs are trained. Then test data can be used to measure the performance of the classifier. For analyzing the performance of a classifier, efficiency is not an accurate metric because that may not be suitably fitted for estimating classification types acquired from imbalanced data sets, i.e. the numbers of observations in different classes differ extensively. So, we can say that for precisely evaluate the performance of a classifier, not only efficiency but also required some other metrics. For a binary decision the confusion matrix, i.e. 2-class problem summarizes the number of true positives (TPs), false positives (FPs), true negatives (TNs), and false negatives (FNs). In the multiclass purpose, i.e. longer than 2-class problems, the resulting confusion matrix dimension is n × n (n > 2). That is known that n is the matrix rows and n is the matrix columns and total entries are n × n. Using that matrix, the number of TPs, FPs, TNs , FNs does not calculate shortly. According to the aforementioned procedure, the values of TPs, FPs, TNs, FNs for class i can be calculated.

Source of Information: International Journal of Engineering and Advanced Technology (IJEAT) Volume-9 Issue-1, October 2019 ISSN: 2249 – 8958,

Article Link (Online Journal):

Funding Source:

Budget:

Total Budget (Tk.) :

Achievement/Findings:

In this paper, we have performed a machine vision-based expert system for recognition of fruit for retrieving our cultural heritage. We have used two feature set which is consists of ten features in total to recognize the fruits. Image processing system has been used to extract all the features. Based on the feature we have able to achieve the desired result. By analyzing the other existing works, we can say that our system works better enough. Our system accuracy level is 94.61% which is good enough. In the future, we will work with a large amount of data of the maximum number of fruits in the context of Bangladesh.

Knowledge Management: Journal