IRJET- Study of Diamond Quality Assessment System using Machine Learning Approach
0 likes12 views
This document presents a study on developing a diamond quality assessment system using machine learning. Currently, diamonds are graded manually which has limitations like low accuracy, subjectivity and inefficiency. The proposed system uses image processing and deep learning concepts to grade diamonds based on their color, texture and clarity. It extracts features from preprocessed diamond images using convolutional neural networks and classifies the diamonds based on the extracted features. This system aims to provide more accurate diamond grading compared to existing manual methods by taking into account different quality aspects through automated analysis of diamond images.
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 07 Issue: 01 | Jan 2020 www.irjet.net p-ISSN: 2395-0072
© 2020, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 678
STUDY OF DIAMOND QUALITY ASSESSMENT SYSTEM USING MACHINE
LEARNING APPROACH
Pratiksha Bhosale1, Vaishnavi Bhosale2,Meet Bikchandani3, Shivani Ghadge4, Prof. Manisha
Navale5
1,2,3,4B.E. Student, Dept. of Computer Engineering, NBN Sinhgad School Of Engineering, Ambegaon, Pune- 411041,
Maharashtra, India
5Professor, Dept. of Computer Engineering, NBN Sinhgad School Of Engineering, Ambegaon, Pune- 411041,
Maharashtra, India
----------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Diamonds for the most part should be cut with
correct shape and extent before they can indicate dazzling
appearance. Precious stones ought to be dull on the off chance
that they are solidified from unadulterated carbon molecules.
At present, jewels are regularly estimated and evaluated by
experienced graders with some extraordinary devices like
amplifying glasses, standard ace stones, colorimeters, ideal
scopes, and so forth. Nonetheless, manual estimation and
evaluating has various disadvantages, for example, restricted
exactness, subjectivity, poor view of respectability, low
proficiency and mind-boggling expense. Therefore to develop
an integrated auto-measuring system for diamond grading is
becoming a challenging research topic. This system presents
methodology for diamond quality grading based on color of
diamond, texture of a diamond and clarity. In order to get
accurate diamond images, a special hardware source is
employed. Quality assessment done through three important
feature extraction of the diamond like color, texture and
clarity. Then extracted features are passed to the classifier for
grading. On the basis of grading quality of a diamond will be
determine.
Key Words: Image Processing, Convolutional neural
network, Pre-processing, Machine learning, Classification.
1. INTRODUCTION
Normal precious stones for the most part should be cut
with correct shape and extent before they can demonstrate
shining appearance. Jewels ought to be boring on the o
chance that they are solidied from unadulterated carbon
particles. On account of a little measure of Nitrogen in most
regular precious stones, jewels regularly demonstrate
different shades of yellow shading. Since common precious
stones regularly demonstrate distinctive shading, different
considerations or then again surrenders, and characteristic
cutting mistakes, they should be re viewed. Other than carat
reviewing, jewel evaluatingincorporatesshadingreviewing,
clearness evaluating, and cut evaluating. At present, jewels
are regularly estimated and evaluated by experienced
graders with some extraordinary devices like amplifying
glasses, standard ace stones, colorimeters, ideal scopes, and
so forth. Nonetheless, manual estimation and evaluatinghas
various disadvantages, for example, restricted exactness,
subjectivity, poor view of respectability, low proficiencyand
mind-boggling expense. Particularly, after ceaselessly
labouring for 60 minutes, the graders eyes regularly
progress toward becoming extremely worn out, and
wronged value acting can be made effortlessly. Therefore to
develop an in tag rated auto-measuring system for diamond
grading is becoming a challenging research topic. Proposed
system works on diamond quality gradingbasedoncolorsof
diamond, texture of a diamond and clarity. In order to get
accurate diamond images, a special hardware source is
employed. Qualityassessmentdonethroughthreeimportant
feature extraction of the diamond like color, texture and
clarity. Then extracted features are passed to the classierfor
grading On the basis of grading quality of a diamond will be
determine.
Fig 1– Diamond color grading
2. LITERATURE SURVEY
[1] Diamond Color Grading Based on Machine Vision
This paper shows a powerful strategy for precious stone
shading reviewing dependent on machine vision. So as to
gain attractive precious stone pictures, an uncommon light
source dependent on a coordinating circle is utilized.
Subsequent to repaying the variance of the light source, the
compositive shading highlights, including free and joint
circulation highlights of Hue andSaturation,are extricatedin
portioned uniform areas. At that point, contingent upon a
prepared BP Neural Network, jewels can be evaluated by
shading. Perceptual Correction for Color GradingofRandom
Textures](https://image.slidesharecdn.com/irjet-v7i1109-200215063533/85/IRJET-Study-of-Diamond-Quality-Assessment-System-using-Machine-Learning-Approach-1-320.jpg)
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 07 Issue: 01 | Jan 2020 www.irjet.net p-ISSN: 2395-0072
© 2020, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 679
[2] Correction for color grading of random textures,
Machine Vision and Applications This paper shows a
powerful strategy for precious stone shading reviewing
dependent on machine vision. Surfaces, the differences of
which are at the edge of human observation. This strategy
utilizes picture rebuilding strategies to recoup an unblurred
form of the picture, and after that obscures it
indistinguishable path from the human visual framework
does, to copy the procedure of the picture being caught by
the human sensor. In this way, the shading picture is
changed into a perceptually uniform shading space, where
shading evaluating happens
3] A Threshold Selection Method from Grey-Level
Histograms A nonparametric and unsupervised technique
for programmed limit determination for picture division is
exhibited. An ideal limit is chosen bythediscriminantrule,to
be specific, in order to boost the detachability of the
resultant classes in dim dimensions. The technique is
exceptionally straightforward, using just the zeroth-and the
main request aggregate snapshots of the dark dimension
histogram. It is clear to stretch out the technique to multi
threshold issues. A few exploratory outcomes are
additionally exhibited to help the legitimacy of the
technique.
[4] Performance analysis ofaColorimeterdesignedwith
RGB color sensor This paper shows a viable technique for
jewel shading evaluating dependentonmachinevision.So as
to secure palatable precious stone pictures, an exceptional
light source dependent on a coordinating circle is utilized. In
the wake of remunerating the vacillation of the light source,
the focused shading highlights, including autonomous and
joint dispersion highlights of Hue and Saturation, are
separated in portioned uniform districts. At that point,
contingent upon a prepared BP Neural Network, jewels can
be evaluated by shading. Examination results demonstrate
that the proposed strategy can achieve an acceptable
precision to substitute manual reviewing for genuine
precious stones. The proposed strategy can likewise be
utilized to group different questions by little shading
distinction
[5] Development of aColorimetricsensorformonitoring
of fish spoilage amines in packaging headspace A
methodological report on essentialness of picturepreparing
and its applications in the field of PC vision is completed
here. Amid a picture preparing Task the informationgivenis
a picture and its yield is an upgraded top notch picture
according to the systems utilized. Picture handling normally
alluded as advanced picture preparing, yet optical and
simple picture handling likewise are conceivable. Our
examination gives a strong prologue to picture handling
alongside division strategies, PC vision essentials and its
connected applications that will be of worth to the picture
preparing and PC vision look into networks.
3. PROPOSED SYSTEM
DIAMOND QUALITY ASSESSMENT
Proposed system
Proposed system works on diamond quality grading based
on color of diamond, texture of a diamond and clarity. In
order to get accurate diamond images, a special hardware
source is employed. Quality assessment done through three
important feature extraction of the diamond like color,
texture and clarity. Then extractedfeaturesarepassedtothe
classifier for grading. On the basis of grading quality of a
diamond will be determine.
Pre-processing –
In image pre-processing, image/picture information
recorded by sensors on a satellite or taken by uncommon
equipment limit mistakes identified with geometry and
splendor estimations of the pixels. These blunders are
amended utilizing suitable scientificmodelswhichare either
distinct or factual models. Picture upgrade is the alteration
of picture by changing the pixel splendor esteemsto enhance
its visual effect. Picture upgrade is the adjustment of picture
by changing the pixel brilliance esteemsto enhanceitsvisual
effect. Picture upgrade includes an accumulation of
procedures that are utilized to enhance the visual
appearance of a picture, or to change over the picture to a
shape which is more qualified for human or machine
elucidation.
Convolutional neural networks - are similar to feed
forward neural networks,wherethe neuronshavelearn-able
weights and biases. Its application have been in signal and
image processing which takes over OpenCV in field of
computer vision.
In this neural network, the input features are taken in batch
wise like a filter. This will help the network to rememberthe
images in parts and can compute the operations. These
computations involve conversion of the image from RGB or
HSI scale to Gray-scale. Once we have this, the changesinthe
pixel value will help detecting the edges and images can be
classified into different categories.
ConvNet are applied in techniques likesignal processingand
image classification techniques. Computervisiontechniques
are dominated by convolutional neural networks because of
their accuracy inimageclassification.Thetechniqueofimage
analysis and recognition, where the agricultureandweather
features are extracted from the open source satellites like
LSAT to predict the future growth and yield of a particular
land are being implemented. In proposed system CNN is
used for image classificationitincludesimagedatasetbelong
to six different classes.](https://image.slidesharecdn.com/irjet-v7i1109-200215063533/85/IRJET-Study-of-Diamond-Quality-Assessment-System-using-Machine-Learning-Approach-2-320.jpg)
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 07 Issue: 01 | Jan 2020 www.irjet.net p-ISSN: 2395-0072
© 2020, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 680
Fig 2: Proposed system architecture diagram
4. PSEUDO CODE
CNN Steps
Step 1: Convolution Operation
The first step in our plan of attack is convolution operation.
In this step, we will focus on feature detectors, which
basically provides as the neural network's filters. We will
also discuss feature maps, understanding the parameters of
such maps, how patterns are detected, the relation re
detected, the layers of detection, and how the findings are
mapped out.
Step 1(b): ReLU Layer
The other section of this step will involve the Rectified
Linear Unit or ReLU. We will take ReLU layers into
consideration and explore how linearity functions in the
context of Convolutional Neural Networks. Notimportant for
understanding CNN's, but there's no harm in a quick lesson
to improve your skills.
Step 2: Pooling
In this part, we'll cover pooling and will get to understand
exactly how it generally works. Our aim here, however, will
be a specific type of pooling; max pooling. We'll cover
different approaches, though, including mean (or sum)
pooling. This part will end with a demonstration made using
a visual interactive tool that will definitely sort the whole
concept out for us.
Step 3: Flattening
This will be a brief breakdown of the flattening process and
how we transform from pooled to flattened layers when
working with Convolutional Neural Networks.
Step 4: Full Connection
In this part, everything that we covered throughout the
section is going to be merged together. By learning this,
you'll get to envision a fuller picture of how Convolutional
Neural Networks operate and how the "neurons" that are
finally produced learn the classification of images.
5. CONCLUSION
This system grades diamond qualityvery effectivelywiththe
image processing techniques and deep learning concepts.
Quality of assessment is increased as compared to the
existing system as all the quality measurements are taken
into the consideration while determining grading like
texture, color and clarity of a diamond.
REFERENCES
1] A. Pacquit, T. L. King, D. Diamond. Development of a
Colorimetric sensor for monitoringoffishspoilageaminesin
packaging headspace. ProceedingsofIEEEonSensors.P.365
- 367 vol.1. 24-27 Oct. 2004.
2] Boukouvalas and M. Petrou. Perceptual correction for
color grading of random textures. Machine Vision and
Applications. Volume 12, P. 129-136, 2000.
3] Diamond Color Grading Based on Machine Vision. Zhiguo
Ren, Jiarui Liao, Lilong Cai, Member, IEEE Department of
Mechanical Engineering, Hong Kong University of Science
and Technology.
4] N. Otsu. A Threshold Selection Method from Gray-Level
Histograms. IEEE Transactions on Systems, Man, and s1976
5] R. A. Sivanantha, K.Sankaranarayanan. Performance
analysis of a Colorimeter designed with RGB color sensor.
ICIAS 2007, International Conference on Intelligent and
Advanced Systems. P. 305-310, 25-28 Nov. 2007.](https://image.slidesharecdn.com/irjet-v7i1109-200215063533/85/IRJET-Study-of-Diamond-Quality-Assessment-System-using-Machine-Learning-Approach-3-320.jpg)
IRJET- Study of Diamond Quality Assessment System using Machine Learning Approach
- 1. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 07 Issue: 01 | Jan 2020 www.irjet.net p-ISSN: 2395-0072 © 2020, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 678 STUDY OF DIAMOND QUALITY ASSESSMENT SYSTEM USING MACHINE LEARNING APPROACH Pratiksha Bhosale1, Vaishnavi Bhosale2,Meet Bikchandani3, Shivani Ghadge4, Prof. Manisha Navale5 1,2,3,4B.E. Student, Dept. of Computer Engineering, NBN Sinhgad School Of Engineering, Ambegaon, Pune- 411041, Maharashtra, India 5Professor, Dept. of Computer Engineering, NBN Sinhgad School Of Engineering, Ambegaon, Pune- 411041, Maharashtra, India ----------------------------------------------------------------------***--------------------------------------------------------------------- Abstract - Diamonds for the most part should be cut with correct shape and extent before they can indicate dazzling appearance. Precious stones ought to be dull on the off chance that they are solidified from unadulterated carbon molecules. At present, jewels are regularly estimated and evaluated by experienced graders with some extraordinary devices like amplifying glasses, standard ace stones, colorimeters, ideal scopes, and so forth. Nonetheless, manual estimation and evaluating has various disadvantages, for example, restricted exactness, subjectivity, poor view of respectability, low proficiency and mind-boggling expense. Therefore to develop an integrated auto-measuring system for diamond grading is becoming a challenging research topic. This system presents methodology for diamond quality grading based on color of diamond, texture of a diamond and clarity. In order to get accurate diamond images, a special hardware source is employed. Quality assessment done through three important feature extraction of the diamond like color, texture and clarity. Then extracted features are passed to the classifier for grading. On the basis of grading quality of a diamond will be determine. Key Words: Image Processing, Convolutional neural network, Pre-processing, Machine learning, Classification. 1. INTRODUCTION Normal precious stones for the most part should be cut with correct shape and extent before they can demonstrate shining appearance. Jewels ought to be boring on the o chance that they are solidied from unadulterated carbon particles. On account of a little measure of Nitrogen in most regular precious stones, jewels regularly demonstrate different shades of yellow shading. Since common precious stones regularly demonstrate distinctive shading, different considerations or then again surrenders, and characteristic cutting mistakes, they should be re viewed. Other than carat reviewing, jewel evaluatingincorporatesshadingreviewing, clearness evaluating, and cut evaluating. At present, jewels are regularly estimated and evaluated by experienced graders with some extraordinary devices like amplifying glasses, standard ace stones, colorimeters, ideal scopes, and so forth. Nonetheless, manual estimation and evaluatinghas various disadvantages, for example, restricted exactness, subjectivity, poor view of respectability, low proficiencyand mind-boggling expense. Particularly, after ceaselessly labouring for 60 minutes, the graders eyes regularly progress toward becoming extremely worn out, and wronged value acting can be made effortlessly. Therefore to develop an in tag rated auto-measuring system for diamond grading is becoming a challenging research topic. Proposed system works on diamond quality gradingbasedoncolorsof diamond, texture of a diamond and clarity. In order to get accurate diamond images, a special hardware source is employed. Qualityassessmentdonethroughthreeimportant feature extraction of the diamond like color, texture and clarity. Then extracted features are passed to the classierfor grading On the basis of grading quality of a diamond will be determine. Fig 1– Diamond color grading 2. LITERATURE SURVEY [1] Diamond Color Grading Based on Machine Vision This paper shows a powerful strategy for precious stone shading reviewing dependent on machine vision. So as to gain attractive precious stone pictures, an uncommon light source dependent on a coordinating circle is utilized. Subsequent to repaying the variance of the light source, the compositive shading highlights, including free and joint circulation highlights of Hue andSaturation,are extricatedin portioned uniform areas. At that point, contingent upon a prepared BP Neural Network, jewels can be evaluated by shading. Perceptual Correction for Color GradingofRandom Textures
- 2. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 07 Issue: 01 | Jan 2020 www.irjet.net p-ISSN: 2395-0072 © 2020, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 679 [2] Correction for color grading of random textures, Machine Vision and Applications This paper shows a powerful strategy for precious stone shading reviewing dependent on machine vision. Surfaces, the differences of which are at the edge of human observation. This strategy utilizes picture rebuilding strategies to recoup an unblurred form of the picture, and after that obscures it indistinguishable path from the human visual framework does, to copy the procedure of the picture being caught by the human sensor. In this way, the shading picture is changed into a perceptually uniform shading space, where shading evaluating happens 3] A Threshold Selection Method from Grey-Level Histograms A nonparametric and unsupervised technique for programmed limit determination for picture division is exhibited. An ideal limit is chosen bythediscriminantrule,to be specific, in order to boost the detachability of the resultant classes in dim dimensions. The technique is exceptionally straightforward, using just the zeroth-and the main request aggregate snapshots of the dark dimension histogram. It is clear to stretch out the technique to multi threshold issues. A few exploratory outcomes are additionally exhibited to help the legitimacy of the technique. [4] Performance analysis ofaColorimeterdesignedwith RGB color sensor This paper shows a viable technique for jewel shading evaluating dependentonmachinevision.So as to secure palatable precious stone pictures, an exceptional light source dependent on a coordinating circle is utilized. In the wake of remunerating the vacillation of the light source, the focused shading highlights, including autonomous and joint dispersion highlights of Hue and Saturation, are separated in portioned uniform districts. At that point, contingent upon a prepared BP Neural Network, jewels can be evaluated by shading. Examination results demonstrate that the proposed strategy can achieve an acceptable precision to substitute manual reviewing for genuine precious stones. The proposed strategy can likewise be utilized to group different questions by little shading distinction [5] Development of aColorimetricsensorformonitoring of fish spoilage amines in packaging headspace A methodological report on essentialness of picturepreparing and its applications in the field of PC vision is completed here. Amid a picture preparing Task the informationgivenis a picture and its yield is an upgraded top notch picture according to the systems utilized. Picture handling normally alluded as advanced picture preparing, yet optical and simple picture handling likewise are conceivable. Our examination gives a strong prologue to picture handling alongside division strategies, PC vision essentials and its connected applications that will be of worth to the picture preparing and PC vision look into networks. 3. PROPOSED SYSTEM DIAMOND QUALITY ASSESSMENT Proposed system Proposed system works on diamond quality grading based on color of diamond, texture of a diamond and clarity. In order to get accurate diamond images, a special hardware source is employed. Quality assessment done through three important feature extraction of the diamond like color, texture and clarity. Then extractedfeaturesarepassedtothe classifier for grading. On the basis of grading quality of a diamond will be determine. Pre-processing – In image pre-processing, image/picture information recorded by sensors on a satellite or taken by uncommon equipment limit mistakes identified with geometry and splendor estimations of the pixels. These blunders are amended utilizing suitable scientificmodelswhichare either distinct or factual models. Picture upgrade is the alteration of picture by changing the pixel splendor esteemsto enhance its visual effect. Picture upgrade is the adjustment of picture by changing the pixel brilliance esteemsto enhanceitsvisual effect. Picture upgrade includes an accumulation of procedures that are utilized to enhance the visual appearance of a picture, or to change over the picture to a shape which is more qualified for human or machine elucidation. Convolutional neural networks - are similar to feed forward neural networks,wherethe neuronshavelearn-able weights and biases. Its application have been in signal and image processing which takes over OpenCV in field of computer vision. In this neural network, the input features are taken in batch wise like a filter. This will help the network to rememberthe images in parts and can compute the operations. These computations involve conversion of the image from RGB or HSI scale to Gray-scale. Once we have this, the changesinthe pixel value will help detecting the edges and images can be classified into different categories. ConvNet are applied in techniques likesignal processingand image classification techniques. Computervisiontechniques are dominated by convolutional neural networks because of their accuracy inimageclassification.Thetechniqueofimage analysis and recognition, where the agricultureandweather features are extracted from the open source satellites like LSAT to predict the future growth and yield of a particular land are being implemented. In proposed system CNN is used for image classificationitincludesimagedatasetbelong to six different classes.
- 3. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 07 Issue: 01 | Jan 2020 www.irjet.net p-ISSN: 2395-0072 © 2020, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 680 Fig 2: Proposed system architecture diagram 4. PSEUDO CODE CNN Steps Step 1: Convolution Operation The first step in our plan of attack is convolution operation. In this step, we will focus on feature detectors, which basically provides as the neural network's filters. We will also discuss feature maps, understanding the parameters of such maps, how patterns are detected, the relation re detected, the layers of detection, and how the findings are mapped out. Step 1(b): ReLU Layer The other section of this step will involve the Rectified Linear Unit or ReLU. We will take ReLU layers into consideration and explore how linearity functions in the context of Convolutional Neural Networks. Notimportant for understanding CNN's, but there's no harm in a quick lesson to improve your skills. Step 2: Pooling In this part, we'll cover pooling and will get to understand exactly how it generally works. Our aim here, however, will be a specific type of pooling; max pooling. We'll cover different approaches, though, including mean (or sum) pooling. This part will end with a demonstration made using a visual interactive tool that will definitely sort the whole concept out for us. Step 3: Flattening This will be a brief breakdown of the flattening process and how we transform from pooled to flattened layers when working with Convolutional Neural Networks. Step 4: Full Connection In this part, everything that we covered throughout the section is going to be merged together. By learning this, you'll get to envision a fuller picture of how Convolutional Neural Networks operate and how the "neurons" that are finally produced learn the classification of images. 5. CONCLUSION This system grades diamond qualityvery effectivelywiththe image processing techniques and deep learning concepts. Quality of assessment is increased as compared to the existing system as all the quality measurements are taken into the consideration while determining grading like texture, color and clarity of a diamond. REFERENCES 1] A. Pacquit, T. L. King, D. Diamond. Development of a Colorimetric sensor for monitoringoffishspoilageaminesin packaging headspace. ProceedingsofIEEEonSensors.P.365 - 367 vol.1. 24-27 Oct. 2004. 2] Boukouvalas and M. Petrou. Perceptual correction for color grading of random textures. Machine Vision and Applications. Volume 12, P. 129-136, 2000. 3] Diamond Color Grading Based on Machine Vision. Zhiguo Ren, Jiarui Liao, Lilong Cai, Member, IEEE Department of Mechanical Engineering, Hong Kong University of Science and Technology. 4] N. Otsu. A Threshold Selection Method from Gray-Level Histograms. IEEE Transactions on Systems, Man, and s1976 5] R. A. Sivanantha, K.Sankaranarayanan. Performance analysis of a Colorimeter designed with RGB color sensor. ICIAS 2007, International Conference on Intelligent and Advanced Systems. P. 305-310, 25-28 Nov. 2007.