Recognizing Characters in Natural Scenes: A Feature Study

Size: px
Start display at page:

Download "Recognizing Characters in Natural Scenes: A Feature Study"

Transcription

1 Recognizing Characters in Natural Scenes: A Feature Study Onur Tekdas tekda001@umn.edu Nikhil Karnad karna015@umn.edu Submitted in partial fulfillment of the course CSci 5521 Pattern Recognition Professor Paul Schrater University of Minnesota, Twin-Cities Fall, 2009 Department of Computer Science and Engineering, University of Minnesota, Twin-Cities

2 Contents 1 Introduction 3 2 Related work 6 3 Features Preprocessing Raw intensities Shape Contexts Wavelet features Principal Component Analysis (PCA) Multiple Kernel Learning 10 5 Results 12 6 Conclusion 14 1

3 List of Figures 1.1 Examples of characters in natural scenes that are difficult to recognize because of the amount of variation Flowchart for a typical character recognition that uses natural scene images as input Characters from the English characters dataset A log-polar histogram for a grayscale image of the digit Histogram for the 0 image shown previously Second level wavelet decomposition with approximation and diagonal coefficients for digit The best performance ( 85%) for WAV E SC112 features with SVM RBF was obtained at kernel width

4 Chapter 1 Introduction A number of factors introduce challenges to the task of recognizing characters in natural scenes. To list a few: 1. Clutter and placement: where exactly in a natural scene is the text and how much of the scene is not relevant to character recognition? 2. Different font styles: outlines vs. solid, thick vs. thin lines, colors and textures, etc. 3. Variation in lighting conditions. Figure 1.1: Examples of characters in natural scenes that are difficult to recognize because of the amount of variation. Figure 1.1 shows the difficulty in recognizing characters from such images. Even if we remove the task of text segmentation, the remaining problems are still formidable. Figure 1.2 shows the typical steps involved with a character recognition system. In this project, we focus on the character recognition aspect, without getting into the details of text segmentation, context, languages etc. We study the effect of different feature vectors on the classification performance for character recognition in natural scenes. Some of the features we tried were: raw grayscale pixel intensities, shape context descriptors, and 3

5 Text Segmentation Character Segmentation Feature Extraction Classification Figure 1.2: Flowchart for a typical character recognition that uses natural scene images as input. wavelet features. Details are explained in Section 3. We use the recent SVM Multiple Kernel Learning (MKL) method [16] for classification because it allows us to optimize weights that linearly combine different kernels to find the best classifier. Our goal in this project was to explore various feature extraction schemes for character recognition, as well as to learn and implement a state-of-the-art classification scheme. Figure 1.3: Characters from the English characters dataset. The dataset was gathered from natural scenes in India for both English and Kannada languages, but we chose to use only the English language characters, as shown in Figure 1.3. Available at 4

6 The authors in [6] also acquired a database of hand-printed characters and another of characters generated by computer fonts, to serve as complementary training data. Upper and lower case characters are treated separately and including digits, there are a total of 62 classes. The dataset has 7705 natural character images. To develop intuition about the features and to work with a manageable data set size, we used only the first 593 data points, which correspond only to digits 0 to 9. 5

7 Chapter 2 Related work Digit and character recoginition is an active area of research in OCR applications, as well as in automatic pattern recognition from natural scenes. In this project, the second type is of relevance. The performance of character recognition largely depends on two main decisions: the feature extraction approach and the classification scheme. First, we study the feature extraction problem by trying out different features that have worked well for researchers in the past. In particular, we borrow features from object recognition literature that are not commonly used in character recognition. For instance, stroke direction is a good feature for numeral recognition [20, 8]. There are a lot of local statistical features that can be used as well. For a detailed survey, the interested reader is referred to [9]. For the classification problem, initial work started with statistical techniques [7] and neural networks [3]. Examples of statistical techniques are: Linear discriminants (LDF), Quadratic discriminants (QDF), Nearest-neighbors (k-nn), Parzen window, etc. Neuralnetworks for character recognition include: Multi-layer perceptron (MLP), Radial basis function network (RBF), polynomial classifier, etc. Details can be found in [3, 11, 18, 13]. In recent years, Support vector machine (SVM) classifiers have gained prominence [4]. It is based on statistical learning theory by Vapnik [14]. An SVM is basically a binary classifier and multiple SVMs can be combined to form a classification system for multi-class classification. The superior performance of SVMs has been justified in numerous experiments, particularly in high dimensionality and small sample size. In the first part of this project, we choose and implement two feature extraction schemes: shape contexts [2] and wavelets [19, 5, 12]. We study their performance using the Spider MATLAB Toolbox [17] for SVM classifiers. In the second part, we implement multiple kernel learning techniques [16, 15, 1] and report our results in comparison to the first part. 6

8 Chapter 3 Features 3.1 Preprocessing We converted all digit images to grayscale and resized them to 30x30 pixels. We used the MATLAB commands rgb2gray and imresize. 3.2 Raw intensities Without any further processing, we use the 593x900 values as a raw dataset. This will henceforth be referred to as RAW. 3.3 Shape Contexts In their paper, Belongie et al [2] describe a method to extract the relative positions of pixels in an edge image (Canny edges). First, we pick 3x3 fixed pixel locations in the image (8 along the border and one in the center). For each location, we impose a log-polar grid and bin the pixels in the edge image into a histogram (see Figures 3.1 and 3.2). We used discretizations: θ = 16 and r = 8. This generates a dataset of size 593x1152, denoted SC1152. We observed that using one fixed pixel location vs. 9 locations only gave a 1% loss in performance, so we opted instead to use 593x112 dataset that we denote as SC112. We implemented this feature extraction scheme in MATLAB. 3.4 Wavelet features Wavelet transforms have been used for texture representation, image compression and character recognition [19, 5, 12]. We resized the images to ( ) and applied a 5-level biorthogonal spline wavelet decomposition. We then retained only the level-2 approximation coefficients (top-left part of Figure 3.3). We denote this dataset as WAV E. We used the 7

9 Figure 3.1: A log-polar histogram for a grayscale image of the digit 0. MATLAB Wavelets Toolbox for this, more specifically the functions wavedec2, appcoef2, detcoef2, wcodemat, wkeep and wdencmp. 3.5 Principal Component Analysis (PCA) For multiple-kernel learning and multi-class svm toolboxes, we frequently encountered an OUT of MEMORY error, even on a computer with 4GB RAM. For these cases, we used PCA as a dimensionality reduction technique, projecting the data points along the top 15 eigenvectors. For the raw dataset, these dimensions capture 66.18% of the data. The command we used for this was MATLAB s princomp command. 8

10 log(r) θ 16 Figure 3.2: Histogram for the 0 image shown previously. Compressed Image Global Threshold = Figure 3.3: Second level wavelet decomposition with approximation and diagonal coefficients for digit 8. 9

11 Chapter 4 Multiple Kernel Learning In this section, we investigate the problem of learning the optimal kernels for SVM. In the previous section, we looked at the problem of choosing the right descriptors for the natural scene characters. There are two important concepts that determines the performance of a descriptor: discriminative power and invariance. The tradeoff between these two concepts changes according the specific problem in hand. In this section, we explored some state of the art kernel learning algorithms, to make this tradeoff automatically. Let N k be a base descriptor and f k be the associated distance function. The descriptor can be expressed in terms of kernels as K k. For example, we can simply set K k (x,x ) = exp(γ k f k (x,x )). Given the base kernels, optimal descriptor s kernel can be approximated as a linear combination of the base kernels, i.e. K opt = k d kk k where d k s correspond to the tradeoff. Then, Multi Kernel Learning (MKL) can be expressed as finding these weights yielding to the optimal kernel. The formulation of MKL is very similar to formulation of ǫsv M, we have learned in the class. 1 Min w,d,η 2 wt w + C1 t ησd t (4.1) subject to y i (w t φ(x i ) + b) 1 η i (4.2) η 0,d 0,Ad p (4.3) where φ t (x i )φ(x j ) = k d k phi t (x i )φ(x j ) (4.4) The only addition to standard ǫsv M formulation is the l 1 regularization of weights d. l 1 regularization causes the sparsity on d. It is desirable to set some of weights to 0 to remove poor descriptors and prevent overfitting [16]. Ad p encodes our prior knowledge about the problem. Using a similar trick in ǫsv M, we can formulate the dual problem as follows: Max α,δ 1 t α + p t δ subject to 1 2 αt Y K k Y α σ k δ t A k 0 δ, 0 α C, 1 t Y α = 0 10

12 General Multi Kernel Learning Benchmark A vs B 91% B vs C 94% B vs D 67% B vs P Error: No support vectors are found! B vs K Error: No support vectors are found! Table 4.1: Attempts that we have made with GMKL toolbox SimpleMKL Benchmark Gaussian kernels (σ 2 ) Polynomial kernels (degree) Result [ ] [1 2 3] 80% [ ] [1 2 3] 75% [ ] [1 2 3] 77% [ ] [1 2 3] 75% [ ] [1 2 3] 80% Table 4.2: Performance of SimpleMKL where non-zero α s correspond to the support vectors, Y is a diagonal matrix of labels and A k is the k th column of A. The dual is convex and it has second order constraints which can be solved as a Second Order Cone Program (SOCP). In [16], Varma et. al. shows how to convert the dual problem into a minmax optimization problem. They claim that the minmax optimization problem can be solved more efficiently and still gives the same result. Varma et. al. published their code [15] on their web site. We used their toolbox (Matlab) to find the optimal kernel for our feature set. This toolbox does not support multi class SVM. Hence, we trained one vs all binary classifiers and used a voting scheme to extend it to multi class problem. The results are listed in Table 4. As seen in Table 4.1, the results from GMKL toolbox is unpredictable. Sometimes accuracy is above 90%, sometimes it is below 70% and occasionally no support vectors are found. We verified by using other algorithms that the reason was not the dataset. Hence, we abandoned using this toolbox. We have found another toolbox called SimpleMKL [1]. SimpleMKL algorithm is presented in [10]. It also provides a multi class svm solution. We have tried a combination of Gaussian and polynomial kernels. The kernels and results are listed in Table 4.2. Since training takes quite awhile (15-30 mins), we were not able to find the optimal parameters in a systematic way. Rather we tried several parameters to manually decide on a good parameter set. Finally we concluded that the best parameters were the ones listed in the first line of the Table 4.2. We used these parameters to compare MKL and standard SVM in Chapter 5. 11

13 Chapter 5 Results We used the Spider machine learning toolbox in MATLAB to measure the 10-fold crossvalidation performance of an SVM RBF classifier on each of our features. The results are shown in the following table. Features # dims Classifier Correct rate RAW 900 SVM rbf= % ± SC SVM rbf= % ± W AV E 196 SVM rbf= % ± W AV E SC SVM rbf= % ± RAW 900 MKL 80% SC MKL % W AV E 196 MKL % W AV E SC MKL % From the results above, we see that using both wavelet features and shape context features together gives us a better performance than using either of them individually, while still getting closer to the performance on the raw data. Even together, both of these features use only about one-third of the dimensions compared to the raw data. This fraction becomes increasingly better for larger images, because we can select an appropriate level of the wavelet approximation coefficients to fit the number of dimensions we would like to have. For the WAV E SC112 feature space, we optimized the width of the RBF kernel in the SVM (see Figure 5.1). However, we found that the best performance is obtained for rbf= with a correct rate of % ± We expected that the Multiple Kernel Learning (MKL) technique would be able to combine these features in an optimal way to achieve better performance, but we frequently encountered an OUT of MEMORY error and therefore had to use PCA to lower the number of dimensions to 15. We also tried Fisher s Linear Discriminant Analysis (LDA) using 12

14 Correct rate RBF kernel width Figure 5.1: The best performance ( 85%) for WAV E SC112 features with SVM RBF was obtained at kernel width fisher command in Spider. The ranking of each component was same. Hence we concluded that we can not decrease the dimensionality using LDA. 13

15 Chapter 6 Conclusion In this project, we considered the digit recognition problem in natural scenes. We started by a literature search. We implemented two common descriptors: shape context and wavelet. We concluded that the recognition performance of shape context is poor while the performance of wawelet is slightly less than the performance of the raw data descriptors. Since the dimension of raw data descriptor is high this result was expected. We used Multiple Kernel Learning to find an optimal kernel for the svm algorithm. However, emprically we showed that these algorithms work poorly in high dimensional feature sets as opposed to the claims in the literature. 14

16 Bibliography [1] Simple multiple kernel learning code. arakotom/- code/mklindex.html. [2] S. Belongie and J. Malik. Matching with shape contexts. In IEEE Workshop on Contentbased Access of Image and Video Libraries, volume 12. Springer, [3] C. Bishop. Neural networks for pattern recognition. Oxford Univ Press, [4] C. Burges. A tutorial on support vector machines for pattern recognition. Data mining and knowledge discovery, 2(2): , [5] G. Chen and T. Bui. Invariant Fourier-wavelet descriptor for pattern recognition. Pattern recognition, 32(7): , [6] T. de Campos, B. Babu, and M. Varma. Character Recognition in Natural Images. In Proceedings of the International Conference on Computer Vision Theory and Applications, Lisbon, Portugal, February [7] F. Keinosuke. Introduction to statistical pattern recognition. Academic Press, Boston, [8] F. Kimura, S. Nishikawa, T. Wakabayashi, Y. Miyake, and T. Tsutsumida. Evaluation and synthesis of feature vectors for handwritten numeral recognition. IEICE Transactions on Information and Systems, 79(5): , [9] C. Liu, K. Nakashima, H. Sako, and H. Fujisawa. Handwritten digit recognition: benchmarking of state-of-the-art techniques. Pattern Recognition, 36(10): , [10] A. Rakotomamonjy, F. Bach, S. Canu, and Y. Grandvalet. SimpleMKL [11] D. Rumelhart, G. Hinton, and R. Williams. Learning representations by backpropagating errors. Cognitive modeling, page 213, [12] D. Shen and H. Ip. Discriminative wavelet shape descriptors for recognition of 2-D patterns. Pattern Recognition, 32(2): ,

17 [13] I. Tarassenko and S. Roberts. Supervised and unsupervised learning in radial basis function classifiers. IEE Proceedings-Vision, Image, and Signal Processing, 141:210, [14] V. Vapnik. The nature of statistical learning theory. Springer Verlag, [15] M. Varma and B. R. Babu. Generalized multiple kernel learning code. [16] M. Varma and D. Ray. Learning the discriminative power-invariance trade-off. In Proc. ICCV, volume Citeseer, [17] J. Weston, A. Elisseeff, G. BakIr, and F. Sinz. The Spider machine learning toolbox [18] D. Wettschereck and T. Dietterich. Improving the performance of radial basis function networks by learning center locations. Advances in neural information processing systems, pages , [19] P. Wunsch and A. Laine. Wavelet descriptors for multiresolution recognition of handprinted characters. Pattern Recognition, 28(8): , [20] M. Yasuda and H. Fujisawa. An improvement of correlation method for character recognition. Trans, IEICE Japan J, 62:

Neural Networks and Deep Learning

Neural Networks and Deep Learning Neural Networks and Deep Learning Example Learning Problem Example Learning Problem Celebrity Faces in the Wild Machine Learning Pipeline Raw data Feature extract. Feature computation Inference: prediction,

More information

FACE RECOGNITION USING SUPPORT VECTOR MACHINES

FACE RECOGNITION USING SUPPORT VECTOR MACHINES FACE RECOGNITION USING SUPPORT VECTOR MACHINES Ashwin Swaminathan ashwins@umd.edu ENEE633: Statistical and Neural Pattern Recognition Instructor : Prof. Rama Chellappa Project 2, Part (b) 1. INTRODUCTION

More information

Fine Classification of Unconstrained Handwritten Persian/Arabic Numerals by Removing Confusion amongst Similar Classes

Fine Classification of Unconstrained Handwritten Persian/Arabic Numerals by Removing Confusion amongst Similar Classes 2009 10th International Conference on Document Analysis and Recognition Fine Classification of Unconstrained Handwritten Persian/Arabic Numerals by Removing Confusion amongst Similar Classes Alireza Alaei

More information

Equation to LaTeX. Abhinav Rastogi, Sevy Harris. I. Introduction. Segmentation.

Equation to LaTeX. Abhinav Rastogi, Sevy Harris. I. Introduction. Segmentation. Equation to LaTeX Abhinav Rastogi, Sevy Harris {arastogi,sharris5}@stanford.edu I. Introduction Copying equations from a pdf file to a LaTeX document can be time consuming because there is no easy way

More information

Support Vector Machines

Support Vector Machines Support Vector Machines RBF-networks Support Vector Machines Good Decision Boundary Optimization Problem Soft margin Hyperplane Non-linear Decision Boundary Kernel-Trick Approximation Accurancy Overtraining

More information

HW2 due on Thursday. Face Recognition: Dimensionality Reduction. Biometrics CSE 190 Lecture 11. Perceptron Revisited: Linear Separators

HW2 due on Thursday. Face Recognition: Dimensionality Reduction. Biometrics CSE 190 Lecture 11. Perceptron Revisited: Linear Separators HW due on Thursday Face Recognition: Dimensionality Reduction Biometrics CSE 190 Lecture 11 CSE190, Winter 010 CSE190, Winter 010 Perceptron Revisited: Linear Separators Binary classification can be viewed

More information

LECTURE 5: DUAL PROBLEMS AND KERNELS. * Most of the slides in this lecture are from

LECTURE 5: DUAL PROBLEMS AND KERNELS. * Most of the slides in this lecture are from LECTURE 5: DUAL PROBLEMS AND KERNELS * Most of the slides in this lecture are from http://www.robots.ox.ac.uk/~az/lectures/ml Optimization Loss function Loss functions SVM review PRIMAL-DUAL PROBLEM Max-min

More information

GENDER CLASSIFICATION USING SUPPORT VECTOR MACHINES

GENDER CLASSIFICATION USING SUPPORT VECTOR MACHINES GENDER CLASSIFICATION USING SUPPORT VECTOR MACHINES Ashwin Swaminathan ashwins@umd.edu ENEE633: Statistical and Neural Pattern Recognition Instructor : Prof. Rama Chellappa Project 2, Part (a) 1. INTRODUCTION

More information

Kernels and Clustering

Kernels and Clustering Kernels and Clustering Robert Platt Northeastern University All slides in this file are adapted from CS188 UC Berkeley Case-Based Learning Non-Separable Data Case-Based Reasoning Classification from similarity

More information

Discriminative classifiers for image recognition

Discriminative classifiers for image recognition Discriminative classifiers for image recognition May 26 th, 2015 Yong Jae Lee UC Davis Outline Last time: window-based generic object detection basic pipeline face detection with boosting as case study

More information

CHAPTER 8 COMPOUND CHARACTER RECOGNITION USING VARIOUS MODELS

CHAPTER 8 COMPOUND CHARACTER RECOGNITION USING VARIOUS MODELS CHAPTER 8 COMPOUND CHARACTER RECOGNITION USING VARIOUS MODELS 8.1 Introduction The recognition systems developed so far were for simple characters comprising of consonants and vowels. But there is one

More information

Use of Multi-category Proximal SVM for Data Set Reduction

Use of Multi-category Proximal SVM for Data Set Reduction Use of Multi-category Proximal SVM for Data Set Reduction S.V.N Vishwanathan and M Narasimha Murty Department of Computer Science and Automation, Indian Institute of Science, Bangalore 560 012, India Abstract.

More information

Feature Extractors. CS 188: Artificial Intelligence Fall Some (Vague) Biology. The Binary Perceptron. Binary Decision Rule.

Feature Extractors. CS 188: Artificial Intelligence Fall Some (Vague) Biology. The Binary Perceptron. Binary Decision Rule. CS 188: Artificial Intelligence Fall 2008 Lecture 24: Perceptrons II 11/24/2008 Dan Klein UC Berkeley Feature Extractors A feature extractor maps inputs to feature vectors Dear Sir. First, I must solicit

More information

Machine Learning for NLP

Machine Learning for NLP Machine Learning for NLP Support Vector Machines Aurélie Herbelot 2018 Centre for Mind/Brain Sciences University of Trento 1 Support Vector Machines: introduction 2 Support Vector Machines (SVMs) SVMs

More information

Introduction to Support Vector Machines

Introduction to Support Vector Machines Introduction to Support Vector Machines CS 536: Machine Learning Littman (Wu, TA) Administration Slides borrowed from Martin Law (from the web). 1 Outline History of support vector machines (SVM) Two classes,

More information

CSE 573: Artificial Intelligence Autumn 2010

CSE 573: Artificial Intelligence Autumn 2010 CSE 573: Artificial Intelligence Autumn 2010 Lecture 16: Machine Learning Topics 12/7/2010 Luke Zettlemoyer Most slides over the course adapted from Dan Klein. 1 Announcements Syllabus revised Machine

More information

CS 343H: Honors AI. Lecture 23: Kernels and clustering 4/15/2014. Kristen Grauman UT Austin

CS 343H: Honors AI. Lecture 23: Kernels and clustering 4/15/2014. Kristen Grauman UT Austin CS 343H: Honors AI Lecture 23: Kernels and clustering 4/15/2014 Kristen Grauman UT Austin Slides courtesy of Dan Klein, except where otherwise noted Announcements Office hours Kim s office hours this week:

More information

Learning to Recognize Faces in Realistic Conditions

Learning to Recognize Faces in Realistic Conditions 000 001 002 003 004 005 006 007 008 009 010 011 012 013 014 015 016 017 018 019 020 021 022 023 024 025 026 027 028 029 030 031 032 033 034 035 036 037 038 039 040 041 042 043 044 045 046 047 048 049 050

More information

Support Vector Machines

Support Vector Machines Support Vector Machines RBF-networks Support Vector Machines Good Decision Boundary Optimization Problem Soft margin Hyperplane Non-linear Decision Boundary Kernel-Trick Approximation Accurancy Overtraining

More information

Classification: Feature Vectors

Classification: Feature Vectors Classification: Feature Vectors Hello, Do you want free printr cartriges? Why pay more when you can get them ABSOLUTELY FREE! Just # free YOUR_NAME MISSPELLED FROM_FRIEND... : : : : 2 0 2 0 PIXEL 7,12

More information

CS 343: Artificial Intelligence

CS 343: Artificial Intelligence CS 343: Artificial Intelligence Kernels and Clustering Prof. Scott Niekum The University of Texas at Austin [These slides based on those of Dan Klein and Pieter Abbeel for CS188 Intro to AI at UC Berkeley.

More information

Machine Learning Final Project

Machine Learning Final Project Machine Learning Final Project Team: hahaha R01942054 林家蓉 R01942068 賴威昇 January 15, 2014 1 Introduction In this project, we are asked to solve a classification problem of Chinese characters. The training

More information

All lecture slides will be available at CSC2515_Winter15.html

All lecture slides will be available at  CSC2515_Winter15.html CSC2515 Fall 2015 Introduc3on to Machine Learning Lecture 9: Support Vector Machines All lecture slides will be available at http://www.cs.toronto.edu/~urtasun/courses/csc2515/ CSC2515_Winter15.html Many

More information

DETECTION OF SMOOTH TEXTURE IN FACIAL IMAGES FOR THE EVALUATION OF UNNATURAL CONTRAST ENHANCEMENT

DETECTION OF SMOOTH TEXTURE IN FACIAL IMAGES FOR THE EVALUATION OF UNNATURAL CONTRAST ENHANCEMENT DETECTION OF SMOOTH TEXTURE IN FACIAL IMAGES FOR THE EVALUATION OF UNNATURAL CONTRAST ENHANCEMENT 1 NUR HALILAH BINTI ISMAIL, 2 SOONG-DER CHEN 1, 2 Department of Graphics and Multimedia, College of Information

More information

HANDWRITTEN GURMUKHI CHARACTER RECOGNITION USING WAVELET TRANSFORMS

HANDWRITTEN GURMUKHI CHARACTER RECOGNITION USING WAVELET TRANSFORMS International Journal of Electronics, Communication & Instrumentation Engineering Research and Development (IJECIERD) ISSN 2249-684X Vol.2, Issue 3 Sep 2012 27-37 TJPRC Pvt. Ltd., HANDWRITTEN GURMUKHI

More information

Feature Extractors. CS 188: Artificial Intelligence Fall Nearest-Neighbor Classification. The Perceptron Update Rule.

Feature Extractors. CS 188: Artificial Intelligence Fall Nearest-Neighbor Classification. The Perceptron Update Rule. CS 188: Artificial Intelligence Fall 2007 Lecture 26: Kernels 11/29/2007 Dan Klein UC Berkeley Feature Extractors A feature extractor maps inputs to feature vectors Dear Sir. First, I must solicit your

More information

Chap.12 Kernel methods [Book, Chap.7]

Chap.12 Kernel methods [Book, Chap.7] Chap.12 Kernel methods [Book, Chap.7] Neural network methods became popular in the mid to late 1980s, but by the mid to late 1990s, kernel methods have also become popular in machine learning. The first

More information

Kernel SVM. Course: Machine Learning MAHDI YAZDIAN-DEHKORDI FALL 2017

Kernel SVM. Course: Machine Learning MAHDI YAZDIAN-DEHKORDI FALL 2017 Kernel SVM Course: MAHDI YAZDIAN-DEHKORDI FALL 2017 1 Outlines SVM Lagrangian Primal & Dual Problem Non-linear SVM & Kernel SVM SVM Advantages Toolboxes 2 SVM Lagrangian Primal/DualProblem 3 SVM LagrangianPrimalProblem

More information

3D Object Recognition using Multiclass SVM-KNN

3D Object Recognition using Multiclass SVM-KNN 3D Object Recognition using Multiclass SVM-KNN R. Muralidharan, C. Chandradekar April 29, 2014 Presented by: Tasadduk Chowdhury Problem We address the problem of recognizing 3D objects based on various

More information

Convolution Neural Networks for Chinese Handwriting Recognition

Convolution Neural Networks for Chinese Handwriting Recognition Convolution Neural Networks for Chinese Handwriting Recognition Xu Chen Stanford University 450 Serra Mall, Stanford, CA 94305 xchen91@stanford.edu Abstract Convolutional neural networks have been proven

More information

Radial Basis Function Neural Network Classifier

Radial Basis Function Neural Network Classifier Recognition of Unconstrained Handwritten Numerals by a Radial Basis Function Neural Network Classifier Hwang, Young-Sup and Bang, Sung-Yang Department of Computer Science & Engineering Pohang University

More information

Robotics Programming Laboratory

Robotics Programming Laboratory Chair of Software Engineering Robotics Programming Laboratory Bertrand Meyer Jiwon Shin Lecture 8: Robot Perception Perception http://pascallin.ecs.soton.ac.uk/challenges/voc/databases.html#caltech car

More information

Mobile Face Recognization

Mobile Face Recognization Mobile Face Recognization CS4670 Final Project Cooper Bills and Jason Yosinski {csb88,jy495}@cornell.edu December 12, 2010 Abstract We created a mobile based system for detecting faces within a picture

More information

Handwritten Hindi Numerals Recognition System

Handwritten Hindi Numerals Recognition System CS365 Project Report Handwritten Hindi Numerals Recognition System Submitted by: Akarshan Sarkar Kritika Singh Project Mentor: Prof. Amitabha Mukerjee 1 Abstract In this project, we consider the problem

More information

Cursive Handwriting Recognition System Using Feature Extraction and Artificial Neural Network

Cursive Handwriting Recognition System Using Feature Extraction and Artificial Neural Network Cursive Handwriting Recognition System Using Feature Extraction and Artificial Neural Network Utkarsh Dwivedi 1, Pranjal Rajput 2, Manish Kumar Sharma 3 1UG Scholar, Dept. of CSE, GCET, Greater Noida,

More information

A Feature based on Encoding the Relative Position of a Point in the Character for Online Handwritten Character Recognition

A Feature based on Encoding the Relative Position of a Point in the Character for Online Handwritten Character Recognition A Feature based on Encoding the Relative Position of a Point in the Character for Online Handwritten Character Recognition Dinesh Mandalapu, Sridhar Murali Krishna HP Laboratories India HPL-2007-109 July

More information

COSC160: Detection and Classification. Jeremy Bolton, PhD Assistant Teaching Professor

COSC160: Detection and Classification. Jeremy Bolton, PhD Assistant Teaching Professor COSC160: Detection and Classification Jeremy Bolton, PhD Assistant Teaching Professor Outline I. Problem I. Strategies II. Features for training III. Using spatial information? IV. Reducing dimensionality

More information

Short Survey on Static Hand Gesture Recognition

Short Survey on Static Hand Gesture Recognition Short Survey on Static Hand Gesture Recognition Huu-Hung Huynh University of Science and Technology The University of Danang, Vietnam Duc-Hoang Vo University of Science and Technology The University of

More information

Content-based image and video analysis. Machine learning

Content-based image and video analysis. Machine learning Content-based image and video analysis Machine learning for multimedia retrieval 04.05.2009 What is machine learning? Some problems are very hard to solve by writing a computer program by hand Almost all

More information

Introduction to object recognition. Slides adapted from Fei-Fei Li, Rob Fergus, Antonio Torralba, and others

Introduction to object recognition. Slides adapted from Fei-Fei Li, Rob Fergus, Antonio Torralba, and others Introduction to object recognition Slides adapted from Fei-Fei Li, Rob Fergus, Antonio Torralba, and others Overview Basic recognition tasks A statistical learning approach Traditional or shallow recognition

More information

Support Vector Machines

Support Vector Machines Support Vector Machines . Importance of SVM SVM is a discriminative method that brings together:. computational learning theory. previously known methods in linear discriminant functions 3. optimization

More information

2. LITERATURE REVIEW

2. LITERATURE REVIEW 2. LITERATURE REVIEW CBIR has come long way before 1990 and very little papers have been published at that time, however the number of papers published since 1997 is increasing. There are many CBIR algorithms

More information

ECG782: Multidimensional Digital Signal Processing

ECG782: Multidimensional Digital Signal Processing ECG782: Multidimensional Digital Signal Processing Object Recognition http://www.ee.unlv.edu/~b1morris/ecg782/ 2 Outline Knowledge Representation Statistical Pattern Recognition Neural Networks Boosting

More information

Robust PDF Table Locator

Robust PDF Table Locator Robust PDF Table Locator December 17, 2016 1 Introduction Data scientists rely on an abundance of tabular data stored in easy-to-machine-read formats like.csv files. Unfortunately, most government records

More information

Shape Context Matching For Efficient OCR

Shape Context Matching For Efficient OCR Matching For Efficient OCR May 14, 2012 Matching For Efficient OCR Table of contents 1 Motivation Background 2 What is a? Matching s Simliarity Measure 3 Matching s via Pyramid Matching Matching For Efficient

More information

Bilevel Sparse Coding

Bilevel Sparse Coding Adobe Research 345 Park Ave, San Jose, CA Mar 15, 2013 Outline 1 2 The learning model The learning algorithm 3 4 Sparse Modeling Many types of sensory data, e.g., images and audio, are in high-dimensional

More information

Traffic Signs Recognition using HP and HOG Descriptors Combined to MLP and SVM Classifiers

Traffic Signs Recognition using HP and HOG Descriptors Combined to MLP and SVM Classifiers Traffic Signs Recognition using HP and HOG Descriptors Combined to MLP and SVM Classifiers A. Salhi, B. Minaoui, M. Fakir, H. Chakib, H. Grimech Faculty of science and Technology Sultan Moulay Slimane

More information

Support Vector Machines

Support Vector Machines Support Vector Machines About the Name... A Support Vector A training sample used to define classification boundaries in SVMs located near class boundaries Support Vector Machines Binary classifiers whose

More information

Preface to the Second Edition. Preface to the First Edition. 1 Introduction 1

Preface to the Second Edition. Preface to the First Edition. 1 Introduction 1 Preface to the Second Edition Preface to the First Edition vii xi 1 Introduction 1 2 Overview of Supervised Learning 9 2.1 Introduction... 9 2.2 Variable Types and Terminology... 9 2.3 Two Simple Approaches

More information

Applications Video Surveillance (On-line or off-line)

Applications Video Surveillance (On-line or off-line) Face Face Recognition: Dimensionality Reduction Biometrics CSE 190-a Lecture 12 CSE190a Fall 06 CSE190a Fall 06 Face Recognition Face is the most common biometric used by humans Applications range from

More information

Classification of Graphical Images in Textbooks

Classification of Graphical Images in Textbooks Classification of Graphical Images in Textbooks Quals Project Report Sangyun Hahn (syhahn@cs.washington.edu) November 24, 2004 Abstract In order to automate the process of transforming figures in textbooks

More information

Class 6 Large-Scale Image Classification

Class 6 Large-Scale Image Classification Class 6 Large-Scale Image Classification Liangliang Cao, March 7, 2013 EECS 6890 Topics in Information Processing Spring 2013, Columbia University http://rogerioferis.com/visualrecognitionandsearch Visual

More information

CS231A Course Project Final Report Sign Language Recognition with Unsupervised Feature Learning

CS231A Course Project Final Report Sign Language Recognition with Unsupervised Feature Learning CS231A Course Project Final Report Sign Language Recognition with Unsupervised Feature Learning Justin Chen Stanford University justinkchen@stanford.edu Abstract This paper focuses on experimenting with

More information

TEXTURE CLASSIFICATION METHODS: A REVIEW

TEXTURE CLASSIFICATION METHODS: A REVIEW TEXTURE CLASSIFICATION METHODS: A REVIEW Ms. Sonal B. Bhandare Prof. Dr. S. M. Kamalapur M.E. Student Associate Professor Deparment of Computer Engineering, Deparment of Computer Engineering, K. K. Wagh

More information

CLASSIFICATION WITH RADIAL BASIS AND PROBABILISTIC NEURAL NETWORKS

CLASSIFICATION WITH RADIAL BASIS AND PROBABILISTIC NEURAL NETWORKS CLASSIFICATION WITH RADIAL BASIS AND PROBABILISTIC NEURAL NETWORKS CHAPTER 4 CLASSIFICATION WITH RADIAL BASIS AND PROBABILISTIC NEURAL NETWORKS 4.1 Introduction Optical character recognition is one of

More information

Handwritten Script Recognition at Block Level

Handwritten Script Recognition at Block Level Chapter 4 Handwritten Script Recognition at Block Level -------------------------------------------------------------------------------------------------------------------------- Optical character recognition

More information

Shape Descriptor using Polar Plot for Shape Recognition.

Shape Descriptor using Polar Plot for Shape Recognition. Shape Descriptor using Polar Plot for Shape Recognition. Brijesh Pillai ECE Graduate Student, Clemson University bpillai@clemson.edu Abstract : This paper presents my work on computing shape models that

More information

Indian Multi-Script Full Pin-code String Recognition for Postal Automation

Indian Multi-Script Full Pin-code String Recognition for Postal Automation 2009 10th International Conference on Document Analysis and Recognition Indian Multi-Script Full Pin-code String Recognition for Postal Automation U. Pal 1, R. K. Roy 1, K. Roy 2 and F. Kimura 3 1 Computer

More information

Comparison of Leaf Recognition using Multi-layer Perceptron and Support Vector Machine

Comparison of Leaf Recognition using Multi-layer Perceptron and Support Vector Machine Comparison of Leaf Recognition using Multi-layer Perceptron and Support Vector Machine [1] Juby George, [2] Gladston Raj S [1] Research Scholar School of Computer Science, Mahatma Gandhi University,Kottayam

More information

Combining Gabor Features: Summing vs.voting in Human Face Recognition *

Combining Gabor Features: Summing vs.voting in Human Face Recognition * Combining Gabor Features: Summing vs.voting in Human Face Recognition * Xiaoyan Mu and Mohamad H. Hassoun Department of Electrical and Computer Engineering Wayne State University Detroit, MI 4822 muxiaoyan@wayne.edu

More information

Supervised Learning (contd) Linear Separation. Mausam (based on slides by UW-AI faculty)

Supervised Learning (contd) Linear Separation. Mausam (based on slides by UW-AI faculty) Supervised Learning (contd) Linear Separation Mausam (based on slides by UW-AI faculty) Images as Vectors Binary handwritten characters Treat an image as a highdimensional vector (e.g., by reading pixel

More information

EE368 Project Report CD Cover Recognition Using Modified SIFT Algorithm

EE368 Project Report CD Cover Recognition Using Modified SIFT Algorithm EE368 Project Report CD Cover Recognition Using Modified SIFT Algorithm Group 1: Mina A. Makar Stanford University mamakar@stanford.edu Abstract In this report, we investigate the application of the Scale-Invariant

More information

Character Recognition from Google Street View Images

Character Recognition from Google Street View Images Character Recognition from Google Street View Images Indian Institute of Technology Course Project Report CS365A By Ritesh Kumar (11602) and Srikant Singh (12729) Under the guidance of Professor Amitabha

More information

Recognition Tools: Support Vector Machines

Recognition Tools: Support Vector Machines CS 2770: Computer Vision Recognition Tools: Support Vector Machines Prof. Adriana Kovashka University of Pittsburgh January 12, 2017 Announcement TA office hours: Tuesday 4pm-6pm Wednesday 10am-12pm Matlab

More information

Sketchable Histograms of Oriented Gradients for Object Detection

Sketchable Histograms of Oriented Gradients for Object Detection Sketchable Histograms of Oriented Gradients for Object Detection No Author Given No Institute Given Abstract. In this paper we investigate a new representation approach for visual object recognition. The

More information

Classification Lecture Notes cse352. Neural Networks. Professor Anita Wasilewska

Classification Lecture Notes cse352. Neural Networks. Professor Anita Wasilewska Classification Lecture Notes cse352 Neural Networks Professor Anita Wasilewska Neural Networks Classification Introduction INPUT: classification data, i.e. it contains an classification (class) attribute

More information

Deep Learning for Computer Vision

Deep Learning for Computer Vision Deep Learning for Computer Vision Spring 2018 http://vllab.ee.ntu.edu.tw/dlcv.html (primary) https://ceiba.ntu.edu.tw/1062dlcv (grade, etc.) FB: DLCV Spring 2018 Yu Chiang Frank Wang 王鈺強, Associate Professor

More information

Facial expression recognition using shape and texture information

Facial expression recognition using shape and texture information 1 Facial expression recognition using shape and texture information I. Kotsia 1 and I. Pitas 1 Aristotle University of Thessaloniki pitas@aiia.csd.auth.gr Department of Informatics Box 451 54124 Thessaloniki,

More information

Support Vector Machines and their Applications

Support Vector Machines and their Applications Purushottam Kar Department of Computer Science and Engineering, Indian Institute of Technology Kanpur. Summer School on Expert Systems And Their Applications, Indian Institute of Information Technology

More information

An Introduction to Content Based Image Retrieval

An Introduction to Content Based Image Retrieval CHAPTER -1 An Introduction to Content Based Image Retrieval 1.1 Introduction With the advancement in internet and multimedia technologies, a huge amount of multimedia data in the form of audio, video and

More information

Robot Learning. There are generally three types of robot learning: Learning from data. Learning by demonstration. Reinforcement learning

Robot Learning. There are generally three types of robot learning: Learning from data. Learning by demonstration. Reinforcement learning Robot Learning 1 General Pipeline 1. Data acquisition (e.g., from 3D sensors) 2. Feature extraction and representation construction 3. Robot learning: e.g., classification (recognition) or clustering (knowledge

More information

Recognition of Handwritten Digits using Machine Learning Techniques

Recognition of Handwritten Digits using Machine Learning Techniques Recognition of Handwritten Digits using Machine Learning Techniques Shobhit Srivastava #1, Sanjana Kalani #2,Umme Hani #3, Sayak Chakraborty #4 Department of Computer Science and Engineering Dayananda

More information

Efficient Tuning of SVM Hyperparameters Using Radius/Margin Bound and Iterative Algorithms

Efficient Tuning of SVM Hyperparameters Using Radius/Margin Bound and Iterative Algorithms IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL. 13, NO. 5, SEPTEMBER 2002 1225 Efficient Tuning of SVM Hyperparameters Using Radius/Margin Bound and Iterative Algorithms S. Sathiya Keerthi Abstract This paper

More information

Machine Learning: Think Big and Parallel

Machine Learning: Think Big and Parallel Day 1 Inderjit S. Dhillon Dept of Computer Science UT Austin CS395T: Topics in Multicore Programming Oct 1, 2013 Outline Scikit-learn: Machine Learning in Python Supervised Learning day1 Regression: Least

More information

Classifying Images with Visual/Textual Cues. By Steven Kappes and Yan Cao

Classifying Images with Visual/Textual Cues. By Steven Kappes and Yan Cao Classifying Images with Visual/Textual Cues By Steven Kappes and Yan Cao Motivation Image search Building large sets of classified images Robotics Background Object recognition is unsolved Deformable shaped

More information

A Taxonomy of Semi-Supervised Learning Algorithms

A Taxonomy of Semi-Supervised Learning Algorithms A Taxonomy of Semi-Supervised Learning Algorithms Olivier Chapelle Max Planck Institute for Biological Cybernetics December 2005 Outline 1 Introduction 2 Generative models 3 Low density separation 4 Graph

More information

Metric learning approaches! for image annotation! and face recognition!

Metric learning approaches! for image annotation! and face recognition! Metric learning approaches! for image annotation! and face recognition! Jakob Verbeek" LEAR Team, INRIA Grenoble, France! Joint work with :"!Matthieu Guillaumin"!!Thomas Mensink"!!!Cordelia Schmid! 1 2

More information

Announcements. CS 188: Artificial Intelligence Spring Classification: Feature Vectors. Classification: Weights. Learning: Binary Perceptron

Announcements. CS 188: Artificial Intelligence Spring Classification: Feature Vectors. Classification: Weights. Learning: Binary Perceptron CS 188: Artificial Intelligence Spring 2010 Lecture 24: Perceptrons and More! 4/20/2010 Announcements W7 due Thursday [that s your last written for the semester!] Project 5 out Thursday Contest running

More information

A Dendrogram. Bioinformatics (Lec 17)

A Dendrogram. Bioinformatics (Lec 17) A Dendrogram 3/15/05 1 Hierarchical Clustering [Johnson, SC, 1967] Given n points in R d, compute the distance between every pair of points While (not done) Pick closest pair of points s i and s j and

More information

SVM Classification in Multiclass Letter Recognition System

SVM Classification in Multiclass Letter Recognition System Global Journal of Computer Science and Technology Software & Data Engineering Volume 13 Issue 9 Version 1.0 Year 2013 Type: Double Blind Peer Reviewed International Research Journal Publisher: Global Journals

More information

Face Recognition Using SIFT- PCA Feature Extraction and SVM Classifier

Face Recognition Using SIFT- PCA Feature Extraction and SVM Classifier IOSR Journal of VLSI and Signal Processing (IOSR-JVSP) Volume 5, Issue 2, Ver. II (Mar. - Apr. 2015), PP 31-35 e-issn: 2319 4200, p-issn No. : 2319 4197 www.iosrjournals.org Face Recognition Using SIFT-

More information

Discovering Visual Hierarchy through Unsupervised Learning Haider Razvi

Discovering Visual Hierarchy through Unsupervised Learning Haider Razvi Discovering Visual Hierarchy through Unsupervised Learning Haider Razvi hrazvi@stanford.edu 1 Introduction: We present a method for discovering visual hierarchy in a set of images. Automatically grouping

More information

LITERATURE REVIEW. For Indian languages most of research work is performed firstly on Devnagari script and secondly on Bangla script.

LITERATURE REVIEW. For Indian languages most of research work is performed firstly on Devnagari script and secondly on Bangla script. LITERATURE REVIEW For Indian languages most of research work is performed firstly on Devnagari script and secondly on Bangla script. The study of recognition for handwritten Devanagari compound character

More information

Software Documentation of the Potential Support Vector Machine

Software Documentation of the Potential Support Vector Machine Software Documentation of the Potential Support Vector Machine Tilman Knebel and Sepp Hochreiter Department of Electrical Engineering and Computer Science Technische Universität Berlin 10587 Berlin, Germany

More information

DEVANAGARI SCRIPT SEPARATION AND RECOGNITION USING MORPHOLOGICAL OPERATIONS AND OPTIMIZED FEATURE EXTRACTION METHODS

DEVANAGARI SCRIPT SEPARATION AND RECOGNITION USING MORPHOLOGICAL OPERATIONS AND OPTIMIZED FEATURE EXTRACTION METHODS DEVANAGARI SCRIPT SEPARATION AND RECOGNITION USING MORPHOLOGICAL OPERATIONS AND OPTIMIZED FEATURE EXTRACTION METHODS Sushilkumar N. Holambe Dr. Ulhas B. Shinde Shrikant D. Mali Persuing PhD at Principal

More information

Facial Expression Recognition using Principal Component Analysis with Singular Value Decomposition

Facial Expression Recognition using Principal Component Analysis with Singular Value Decomposition ISSN: 2321-7782 (Online) Volume 1, Issue 6, November 2013 International Journal of Advance Research in Computer Science and Management Studies Research Paper Available online at: www.ijarcsms.com Facial

More information

ECE 285 Class Project Report

ECE 285 Class Project Report ECE 285 Class Project Report Based on Source localization in an ocean waveguide using supervised machine learning Yiwen Gong ( yig122@eng.ucsd.edu), Yu Chai( yuc385@eng.ucsd.edu ), Yifeng Bu( ybu@eng.ucsd.edu

More information

Isolated Curved Gurmukhi Character Recognition Using Projection of Gradient

Isolated Curved Gurmukhi Character Recognition Using Projection of Gradient International Journal of Computational Intelligence Research ISSN 0973-1873 Volume 13, Number 6 (2017), pp. 1387-1396 Research India Publications http://www.ripublication.com Isolated Curved Gurmukhi Character

More information

A Real Time Facial Expression Classification System Using Local Binary Patterns

A Real Time Facial Expression Classification System Using Local Binary Patterns A Real Time Facial Expression Classification System Using Local Binary Patterns S L Happy, Anjith George, and Aurobinda Routray Department of Electrical Engineering, IIT Kharagpur, India Abstract Facial

More information

Skin and Face Detection

Skin and Face Detection Skin and Face Detection Linda Shapiro EE/CSE 576 1 What s Coming 1. Review of Bakic flesh detector 2. Fleck and Forsyth flesh detector 3. Details of Rowley face detector 4. Review of the basic AdaBoost

More information

Image Segmentation. Srikumar Ramalingam School of Computing University of Utah. Slides borrowed from Ross Whitaker

Image Segmentation. Srikumar Ramalingam School of Computing University of Utah. Slides borrowed from Ross Whitaker Image Segmentation Srikumar Ramalingam School of Computing University of Utah Slides borrowed from Ross Whitaker Segmentation Semantic Segmentation Indoor layout estimation What is Segmentation? Partitioning

More information

ROBUST SCENE CLASSIFICATION BY GIST WITH ANGULAR RADIAL PARTITIONING. Wei Liu, Serkan Kiranyaz and Moncef Gabbouj

ROBUST SCENE CLASSIFICATION BY GIST WITH ANGULAR RADIAL PARTITIONING. Wei Liu, Serkan Kiranyaz and Moncef Gabbouj Proceedings of the 5th International Symposium on Communications, Control and Signal Processing, ISCCSP 2012, Rome, Italy, 2-4 May 2012 ROBUST SCENE CLASSIFICATION BY GIST WITH ANGULAR RADIAL PARTITIONING

More information

Discriminate Analysis

Discriminate Analysis Discriminate Analysis Outline Introduction Linear Discriminant Analysis Examples 1 Introduction What is Discriminant Analysis? Statistical technique to classify objects into mutually exclusive and exhaustive

More information

Conspicuous Character Patterns

Conspicuous Character Patterns Conspicuous Character Patterns Seiichi Uchida Kyushu Univ., Japan Ryoji Hattori Masakazu Iwamura Kyushu Univ., Japan Osaka Pref. Univ., Japan Koichi Kise Osaka Pref. Univ., Japan Shinichiro Omachi Tohoku

More information

An Approach for Reduction of Rain Streaks from a Single Image

An Approach for Reduction of Rain Streaks from a Single Image An Approach for Reduction of Rain Streaks from a Single Image Vijayakumar Majjagi 1, Netravati U M 2 1 4 th Semester, M. Tech, Digital Electronics, Department of Electronics and Communication G M Institute

More information

Semi-Automatic Transcription Tool for Ancient Manuscripts

Semi-Automatic Transcription Tool for Ancient Manuscripts The Venice Atlas A Digital Humanities atlas project by DH101 EPFL Students Semi-Automatic Transcription Tool for Ancient Manuscripts In this article, we investigate various techniques from the fields of

More information

Classification Algorithm for Road Surface Condition

Classification Algorithm for Road Surface Condition IJCSNS International Journal of Computer Science and Network Security, VOL.4 No., January 04 Classification Algorithm for Road Surface Condition Hun-Jun Yang, Hyeok Jang, Jong-Wook Kang and Dong-Seok Jeong,

More information

Applying Supervised Learning

Applying Supervised Learning Applying Supervised Learning When to Consider Supervised Learning A supervised learning algorithm takes a known set of input data (the training set) and known responses to the data (output), and trains

More information

Bagging for One-Class Learning

Bagging for One-Class Learning Bagging for One-Class Learning David Kamm December 13, 2008 1 Introduction Consider the following outlier detection problem: suppose you are given an unlabeled data set and make the assumptions that one

More information

Feature scaling in support vector data description

Feature scaling in support vector data description Feature scaling in support vector data description P. Juszczak, D.M.J. Tax, R.P.W. Duin Pattern Recognition Group, Department of Applied Physics, Faculty of Applied Sciences, Delft University of Technology,

More information