APPLICATION OF MACHINE LEARNING - AUTOMATED FRUIT SORTING TECHNIQUE B.Anudeep V. Gowtham Chandra (102U1A0503) (102U1A05537) IVth CSE GEETHANJALI INSTITUTE OF SCIENCE & TECHNOLOGY NELLORE anudeep.badam@gmail.com gowtham.viper9@gmail.com ABSTRACT: Finally Automated Fruit sorting is latest trend and efficient technique in modern Machine learning is one of the discipline in Agriculture. Data Mining. Machine learning and data mining are relatively similar. But the machine learning focuses on prediction, , Keywords: based on known properties learned from training data. In this we have explained Machine learning, Data mining, supervised about classification of machine learning learning, unsupervised learning, reinforcement algorithms, list of machine learning learning, semi-supervised learning, computer algorithms and applications of machine vision, image processing, KNN, LDA,SVM algorithms, Transduction, Induction. learning. In addition to we have explained about APPLICATION OF MACHINE Introduction LEARNING-FRUIT SORTING TECHNIQUE. Using image processing, computer vision, Machine learning is one of the discipline in data mining.It is a “Field of study that gives computers the machine learning techniques we have ability to learn without being explicitly done fruit sorting. Based on quality of fruit programmed”. with data/parameters CLASSIFICATION is “A computer program is said to learn from done using machine learning algorithms experience E with respect to some class of tasks T like Support vector measure (SVM ), K- and performance measure P,if its performance at Nearest neighbor (KNN),fuzzy logic tasks in T,as measured by P,improves with experience E” approach, Artificial Neural networks. “A core objective of learner is to GENERALIZE Unsupervised Learning: from EXPERIENCE”. In this algorithms operate on unlabeled Generalization means ability to learn the machine to examples i.e, input where the desired output perform accurately on new,unseen tasks after having is unknown.in this main objective is to experienced a learning data set. discover structure in data(Cluster analysis). Machine learning Vs Data Mining: Semi-supervised Learning: Many of them confuse about data mining and In this it combines both labelled and machine learning that both are same. But the terms unlabelled examplesto generate a function or are different . classifier. Transductive Learning: What is Machine learning? It predicts new outputs on specific,fixed test cases and training cases. Machine learning focuses on Prediction, based on known properties learned from training data. Reinforcement Learning: In this agent attempts to gather knowledge What is Data mining? from environment.The agents executes actions which cause the observable state of Data mining focuses on discovery of unknown the environment to change. properties in the data.It is also known as Knowledge Inductive Learning: discovery in data base. It is learning from based on previous experience. Why we use Machine learning? Developmental Learning: Machine learning systems attempt to eliminate the This is latest Learning known as Robot need for human intuition in data analysis. However learning. some human interaction can be reduced. List of Machine Learning Classification of algorithms in Machine Algorithms: learning:  Support vector machines  Artificial neural networks  Supervised Learning  Clustering  Unsupervised Learning  Decision tree learning  Semi-supervised Learning  Association rule learning  Transduction  Bayesian networks  Reinforcement Learning  Metric learning algorithms  Inductive Learning  Developmental Learning Applications of Machine learning:  Speech recognition Supervised Learning:  Drive automobiles In this algorithms are trained on labelled  Play world-class backgammon examples, i.e, inputs where the desired  Program generation outputs is known. supervised learning  Routing in communication networks algorithms attempts to generalize a function  Understanding handwritten text or mapping from inputs to outputs.  Data mining   face detection and face recogniastion Computer vision systems provide rapid, economic,  Automated fruit sortiong technique hygienic, consistent and objective assessment. Diffi culties still exist in this fi eld due to relativity slow Automated Fruit Sorting Application commercial uptake of computer vision technology and processing speeds still fail to meet modern In general farmers and distributors do conventional manufacturing requirements in all sectors. A model quality inspection and handpicking to sort and grade for sorting is proposed in order to overcome agricultural and food products. drawback of current grading systems. so we use This manual method is time-consuming, laborious, machine learning techniques. less efficient, monotonous, slow and inconsistent. Parameters considered in Computer Automated fruit sorting is done Using techniques vision: like :  Image Processing  Computer Vision  Machine Learning Advantages of Automated fruit sorting:  Cost effective.  Consistent  Greater product stability.  Safety.  Speed and accurate sorting can be achieved.  Improve the quality of the product.  Abolish inconsistent manual evaluation. Drawbacks in computer vision:  Reduce dependence on available • Current sorting systems are not accurate. traditional inspection. • Very few parameters like size and color are Quality of fruit sorting considered for grading systems. Parameters: • Still all are under research laboratories. Flavor such as sweetness, acidity content in the • Most of research and development of automated product, grading through appearance on bases agriculture product sorting has been done outside color,size, shape, blemishes and glossiness of India. product, and texture that is assorted on its firmness or product’s mouth feel. Below tables summarize some • The sorting of fruits is still performed manually in of the very recent grading and sorting systems. India. COMPUTER VISION • No grading system is yet available for fruits like chikoo, sugarcane and grapes etc that are exported to other countries from India Proposed Model network approach are the examples of segmentation methods.  Size parameter: From the segmented image, size parameter can be identified using machine vision by measuring projected area, perimeter or diameter.  Shape parameter : can be identified using contour based methods like chain code, B-spine, Hausdorff distance ,Fourier descriptor, etc. or region based methods like convex hull, medial axis, Legendre moments, shape matrix etc.  color parameter: On moving up to next, color feature can be identified using color features of fruits and vegetables included mean, variance, ranges of the red, green, and blue color primaries Fig: Proposed model for automated fruit (RGB color model) and the derived hue, saturation, and intensity values (HIS color sorting model). As shown in figure, first fruits are collected in  Skin texture parameter: a chamber. From the chamber it moves through Skin disease and defects can be found out escalators safely where the weight of the fruit gets using skin texture identification methods. estimated. It moves towards another chamber where  Interior appearance parameter: the image of fruit is captured by more than one Image descriptors like global color camera in different angle. histogram; Unser’s descriptors, color coherence vectors, border/interior, Detecting fruit growth: appearance descriptors etc. can be used for classification of fruits and vegetables. For detecting fruit growth (raw or ripped), smell of the fruit is detected by sensors of wireless sensor  MACHINE LEARNING ALGORITHMS network. USED: Finally, machine-learning algorithm is used Usage of image processing: for classification of parameters. Machine learning algorithms are neural network, Image is then processed where various algorithms are fuzzy logic, genetic algorithm, fractal applied on image for finding expected features like dimensions, Support Vector Machine size, depth, 3D model, texture and color. For finding (SVM), K- Nearest Neighbor (KNN), Linear different features of fruit image following steps Discriminant Analysis (LDA) etc. should be applied, Based on the decision drawn after the process on the above steps, the fruit is  Image segmentation algorithm can be classified into different categories like big, applied on captured image.Histogram small, medium sized, ripe/unripe or thresolding, feature space clustering, Region defectives. basedapproach; Edge detection approach,fuzzy approach and neural  Finally automatic packaging system packs [4] Tajul Rosli Bin Razak, Mahmod Bin Othman the fruit according to the categories (DR), Mohd Nazari Bin Abu Bakar (DR), Khairul provided. Adilah BT Ahmad, and AB.Razak Bin Mansor, "Mango Grading By Using Fuzzy Image Analysis," FUTURE APPLICATIONS: In proceedings of International Conference on Agricultural, Environment and Biological Sciences, o Rice Sorting Phuket, 2012. o Paddy Sorting o Pulses Sorting [5] http://www.ibef.org/industry/agriculture- india. [6] Xu Liming and Zhao Yanchao, "Automated Conclusion and Future Direction: strawberry grading system based on image processing," Computers and Electronics in Agriculture, vol. 71, no. Supplement 1, pp. S32-S39,  Automated fruit sorting is April 2010. speedy, inexpensive, safe and accurate.  Proposed model is generalized and it is considering far more feature parameters than available sorting systems.  Currently, research in the automated fruit sorting and grading has been conducted by experimenting them in laboratories only. References: [1] Dah-Jye Lee, James K. Archibald, and Guangming Xiong, "Rapid Color Grading for Fruit Quality Evaluation Using DirectColor Mapping," IEEE TRANSACTIONSON AUTOMATION SCIENCE AND ENGINEERING, vol. 8, no. 2, pp. 292-302,November 2011. [2] Mahendran R, Jayashree GC, Alagusundaram K, "Application of Computer Vision Technique on Sorting and Grading of Fruits and Vegetables ", Abd El-Salam et al., J Food Process Technology 2011. [3] Pib.nic.in/archieve/others/2012/mar/ d2012031302.pdf, Date: 11.7.2013.