Nothing Special   »   [go: up one dir, main page]
Scribd Logo
Upload

Welcome to Scribd!

  • 61 views

    Image Classification (Project Introducing)

    Uploaded by

    DMadi Asriny
    This document discusses image classification using convolutional neural networks (CNN). It provides examples of CNN achieving high accuracy rates for facial recognition (89%) and temple image classification (98.99% training, 85.57% test). The architecture of CNN is divided into feature extraction and fully-connected layers. The author then discusses their project using CNN for vegetable/fruit classification to make the selection process easier and less time-consuming than manual methods. Their methodology involves collecting image data, preprocessing, processing with CNN, and analyzing histograms and deep learning methods.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd

    Image Classification (Project Introducing)

    Uploaded by

    DMadi Asriny
    61 views8 pages
    This document discusses image classification using convolutional neural networks (CNN). It provides examples of CNN achieving high accuracy rates for facial recognition (89%) and temple image classification (98.99% training, 85.57% test). The architecture of CNN is divided into feature extraction and fully-connected layers. The author then discusses their project using CNN for vegetable/fruit classification to make the selection process easier and less time-consuming than manual methods. Their methodology involves collecting image data, preprocessing, processing with CNN, and analyzing histograms and deep learning methods.

    Original Description:

    Project Introducing

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document discusses image classification using convolutional neural networks (CNN). It provides examples of CNN achieving high accuracy rates for facial recognition (89%) and temple image classification (98.99% training, 85.57% test). The architecture of CNN is divided into feature extraction and fully-connected layers. The author then discusses their project using CNN for vegetable/fruit classification to make the selection process easier and less time-consuming than manual methods. Their methodology involves collecting image data, preprocessing, processing with CNN, and analyzing histograms and deep learning methods.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    61 views8 pages

    Image Classification (Project Introducing)

    Uploaded by

    DMadi Asriny
    This document discusses image classification using convolutional neural networks (CNN). It provides examples of CNN achieving high accuracy rates for facial recognition (89%) and temple image classification (98.99% training, 85.57% test). The architecture of CNN is divided into feature extraction and fully-connected layers. The author then discusses their project using CNN for vegetable/fruit classification to make the selection process easier and less time-consuming than manual methods. Their methodology involves collecting image data, preprocessing, processing with CNN, and analyzing histograms and deep learning methods.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    of 8

    Image Classification

    by : Dhiya Mahdi Asriny


    Convolutional Neural Networks Method
    Method of CNN is one of the Deep Learning method that
    is being developed at the moment. Network on CNN are
    made with the assumption that the insert is used in the
    form of pictures. This network has a special coating which
    is named with a layer of convolution, where on this layer
    an image insert will be processed based on the filters that
    are already specified. Of each layer it will produce a
    pattern of some parts of the image that will be easier to
    be classified, making learning proccess more efficient to
    be implemented.

    Some research on image processing with CNN method get good accuracy results :
    • Muhammad Zufar and Budi Setiyono (2016) for facial recognition in real-time. The accuracy of
    the results obtained, namely of 89%.
    • Kevin Danukusumo Pudi (2017) Temple-based image classification for the GPU. Optimal test
    results against the image of the temple shows the accuracy of 98.99% on training sets and
    85.57% on the test set. The study mentions that the technique of Deep Learning with CNN
    was able to do image classification Temple very well.
    The architecture of CNN is divided into 2 major parts, the Extraction Layer Feature and the Fully-Connected Layer
    (MLP). I use this term because the process that occurs in this section is "encoding" an image into features in the form of
    numbers that represent the image (Feature Extraction). Feature extraction layer consists of two parts. Convolutional Layer and
    Merge Layer.
    ABOUT MY PROJECT

    CLASSIFICATION
    The Development of Technology Convolutional Neural Networks

    Process selection vegetables or fruits generally only done manually and involve
    human as decision makers. It has several weaknesses which are :
    • The time it takes relatively long
    • Humans also tend to feel tired and easily feel bored if doing a monotonous activity
    • Difference perception about the quality of the fruit
    • It can also result in inconsistent in election process, because influenced by psychic
    condition of human
    • Manually way done too much time consuming, so if applied on a large industrial scale, it
    required help engine in the process.
    FRUITS OR VEGETABLES
    CLASSIFICATION
    As long as this is the case, I want to make a classification
    for fruits or vegetables to make easy for human work. To
    see which quality good and bad.
    RESEARCH METHODOLOGY
    • Type and Source of Data
    Type of data used in this research is the primary data. The data is obtained by taking the
    image directly using camera smartphone, for more accurate data.

    • Method of Data Analisys


    1. Histogram of image that is used to view a representation of colors distribution from an image.
    2. Deep Learning Method that is Convolutional Neural Networks that is used to classify the image.

    • Research Steps
    1. Data Collection
    2. Image Prepocessing
    3. Image Processing
    Thank You!

    You might also like