Combining Selective Search Segmentation and Random Forest for Image Classification

Size: px
Start display at page:

Download "Combining Selective Search Segmentation and Random Forest for Image Classification"

Transcription

1 Combining Selective Search Segmentation and Random Forest for Image Classification Gediminas Bertasius November 24, Problem Statement Random Forest algorithm have been successfully used in many computer vision tasks such as image classification [1] and image segmentation [4]. Recently, Yao et al. showed that a random forest composed of the decision trees where every node is a discriminative classifier outperforms state-of-the-art results in the finegrained image categorization problems [8]. Yao et al. attributed their success to the two main components of their system: discrimination and randomization. Discrimination refers to the use of SVM to learn the splits at each node, whereas randomization refers to a random selection of image patches, which are used as a form of features to learn the splits at each node. There are several problems that may arise from this randomization procedure. Firstly, if we consider image patches of size in an image, sampling space may contain thousands of patches, which makes it less likely that a randomly selected patch will contain an object of interest for the image categorization. In addition, randomly selected samples are more likely to overlap with each other, which would cause redundancy. Therefore, in this project, I investigated new ways for selecting image patches. In theory, more informative patch selection should result in higher quality splits at each tree node, which in turn should increase overall accuracy of the classifier. Figure 1: image a) illustrates how random image patch selection might work in original Yao s et al. [8] algorithm. Image b) shows the regions that we should be using instead, to learn the splits in a tree growing procedure. 1

2 2 Proposed Method Specifically, to fix the problems related to random patch selection I integrated a selective search segmentation algorithm [5] into the original random forest framework. Image patches selected using selective search segmentation are more likely to contain the objects of interest. In addition, segmentation should eliminate redundant overlapping between the image patches, which will make our feature space more diverse. Fixing these two problems should result in an increased discriminative power of random forest. My proposed method works in a following way. First, for each tree a random subset of images is selected for the training stage. Then, selective search segmentation algorithm is used on these images to produce coordinates of the most relevant regions in each image. Afterwards, at each split the algorithm randomly picks N regions from the regions produced by selective search segmentation. Then, the features from these regions are trained via SVM model and the region that produces the best split is selected as a splitting rule (splitting rule would be represented in a form of region coordinates). All of the images are normalized (resized) beforehand so that the coordinates would correspond to the same region in each image. Below is a pseudocde that describes high-level procedure of growing a decision tree according to my proposed method: for each tree t do do -Select a set of training examples D; -Randomly pick a subset of coordinates corresponding to image patches produced by a selective search segmentation; -From each image extract image patches corresponding to the selected coordinates; -Train SVM on each image patch and then select the best image patch to split the dataset D into D 1, D 2 ; -Recursively split datasets D 1, D 2 ; -Return tree t; end Algorithm 1: Proposed Method for Tree Growing Procedure 3 Implementation 3.1 Low Level Features Similarly to [8] I used SIFT [3] visual descriptors as my low level image representation. After extracting SIFT descriptors, I applied k-means clustering algorithm to construct visual vocabularies of size 1024 and 256 for Caltech 256 and Stanford 40 Actions datasets respectively. Then I utilized Locality Constrained Linear Coding [6] to match the descriptors with the specific words in the constructed vocabulary. 3.2 Selective Search Segmentation Before beginning Random Forest procedure, I standardize each image by rescaling them to the same size and then apply Selective Search Segmentation to extract important regions from each image. Each region is represented by 4 coordinates in the image (points in the bottom left and top right corners of the 2

3 region). Then, k-means is applied to all the regions that were returned by Selective Search Segmentation and its centroids are chosen as the final candidate regions. In this particular case, I used 1024 centroids. 3.3 Decision Tree Framework To build the trees, I use an identical scheme as in [8]. However, instead of choosing candidate image regions randomly as is done in [8] I choose from the image regions that were previously selected by selective search segmentation algorithm (as described in section 3.2). At every node, each of the selected regions is considered as a possible splitting rule for the node, where the region in the tree is represented by its 4 coordinates. For each region, I use linear SVM to determine the best hyperplane for splitting the remaining images at that node. Then the region and its respective SVM model that had highest information gain are stored as splitting rules at that particular node. It is important to note that before applying SVM the labels of the remaining images are randomly binarized in a way that images belonging to the same class share the same binary label. This procedure allows to apply binary SVM procedure rather than having to use the multi-class SVM, which is highly desirable. 4 Evaluation 4.1 Datasets To compare the performance of my proposed method with the original Yao s et al. [8] algorithm, I ran both methods on two datasets: Caltech 256 [2] and Stanford 40 Actions dataset [7]. Caltech 256 contains 256 image categories and have approximately 90 image samples for each category. Stanford 40 Actions dataset consists of images of humans performing 40 different actions where each action contains 300 images. It is important to note that these two datasets are significantly different. In Caltech 256, most of the objects are localized and centered at the image. In addition, there is minimal background clutter in the images. Stanford 40 Actions dataset, on the other hand, is completely opposite in these aspects. Actions of interest may appear anywhere in the image, not necessarily in the center. In addition, most of the images in Stanford 40 Actions dataset contain other objects and lots of background clutter, which makes classification significantly more challenging. Sample images from both datasets are presented in Figure Results Results suggest that incorporating Selective Search Segmentation algorithm into Yao s et al. [8] Random Forest framework yields higher accuracy rates when classification is done on challenging image datasets, in which objects are not localized and there is lots of background clutter. All of this makes sense and perfectly aligns with my expectations. Obviously if the objects in the images are well localized and there is no background clutter, segmentation is simply redundant because almost every patch in the image is informative by itself. However, in the case where images contain other objects in the background, segmentation helps to identify which of those objects are actually relevant for 3

4 Figure 2: a) illustrates types of images in Caltech 256 dataset (object class is baseball glove in this case) whereas b) displays a sample image from Stanford 40 Actions dataset (action class is washing dishes in this case) the classification, which increases the accuracy of the classifier. All of these hypotheses are well supported by the results presented in the sections below Results on Stanford 40 Actions Dataset Just as discussed earlier, incorporating segmentation into the original algorithm yields higher accuracy on Stanford 40 Actions dataset. I compared both of the methods for different number of samples used in the training procedure, and in all cases my proposed method produced higher accuracy rates. Analyzing accuracy rates for individual classes also reveal that selecting patches via segmentation is beneficial for specific class identification. My proposed method performed better in all but one classes as illustrated in Figure 3. Figure 3: plot a) shows accuracy of the two methods for different number of training samples whereas b) displays the accuracies of both methods for sample of individual classes in Stanford 40 Actions dataset in Figure 4, I also presented the confusion matrices produced by both methods. The summary of the results for Stanford 40 Actions dataset are presented in Figure 5. As a side note, it is interesting to note that the accuracy rate in Figure 3 is not steadily improving as we increase the number of samples used in the 4

5 training procedure. This can be explained by a couple of things. First, with the higher number of samples it is clearly much harder for the linear SVM to learn meaningful splits. Furthermore, I did not have enough time to fine tune all of the decision tree parameters. Therefore, with the higher number of training samples the classifier may be slightly overfitting. These couple of details would explain the accuracy behavior in Figure 3. Figure 4: Confusion matrices produced by original Yao s et al. [8] algorithm and my modified algorithm respectively on Stanford 40 Actions Figure 5: Summary of the results of both methods on Stanford 40 Actions dataset Results on Caltech 256 Dataset However, incorporating segmentation into Yao s et al. original algorithm did not improve success rates on Caltech 256 dataset. As mentioned earlier, this is because images in Caltech 256 dataset are already well localized and do not contain much background clutter. As a result, it is possible to select very informative patches even via random patch selection. In fact, random patch selection in this case may be even more beneficial as it produces more diverse trees, which in turn improves accuracy of the overall classification. These statements are well supported by the Figure 6, which illustrates that random patch selection produces equivalent or even better results on Caltech 256. In Figure 7 I also presented confusion matrices produced by both methods. In addition, the summary of the results for Caltech 256 dataset are presented in Figure 8. Similarly to the results for Stanford 40 Actions dataset, the accuracy rate in Figure 6 does not exhibit a steady increase with the higher number of training examples. Once again this could be explained by the same reasons: linear SVM may fail to find good splits with the higher number of samples. Furthermore due to suboptimal learning parameters decision trees may be overfitting the data as I increase the number of training samples. 5

6 Figure 6: plot a) shows accuracy of the two methods for different number of training samples whereas b) displays the accuracies of both methods for sample of individual classes in Caltech 256 dataset Figure 7: Confusion matrices produced by original Yao s et al. [8] algorithm and my modified algorithm respectively on Caltech 256 Figure 8: Summary of the results of both methods on Caltech 256 dataset 5 Conclusions and Future Work As illustrated by the results, it is only beneficial to incorporate segmentation into the original Yao s et al. algorithm in the case when images in the dataset are very challenging. That includes cases when images contain lots of background objects and when there is no localization of the objects of interest. In such cases segmentation will definitely help to identify which regions in the image are important for the classification. In addition, because segmentation can be applied before the training procedure this will not affect the run time of decision tree learning procedure in any way, a property which is highly desirable. However, in the cases when images are well localized and do not contain any background clutter (like images in Caltech 256), picking regions via segmentation may actually hurt the performance of the classifier. This is because the 6

7 resulting trees will be less diverse, which is an important criteria for the overall effectiveness of a random forest framework. Overall, I believe my proposed method is highly beneficial because most of the images in the real word will indeed contain lots of background objects, in which case the regular Yao s et al. algorithm may not perform as well as my proposed method. Having said that, I believe some improvements could be made to enhance the performance of this algorithm. For one, there may be a better way to represent patches within the trees than simply storing them as coordinates in an image. This coordinate representation makes an implicit assumption that objects are distributed in a uniform way across all of the images, which is obviously not a correct assumption in most cases. Therefore, better ways to represent patches inside the tree structure should be explored in the future. References [1] A. Bosch, A. Zisserman, and X. Munoz. Image classification using random forests and ferns. In IEEE International Conference on Computer Vision, [2] G. Griffin, A. Holub, and P. Perona. Caltech-256 Object Category Dataset. Technical Report CNS-TR , California Institute of Technology, [3] David G. Lowe. Distinctive image features from scale-invariant keypoints. International Journal of Computer Vision, 60:91 110, [4] Jamie Shotton, Matthew Johnson, and Roberto Cipolla. Semantic texton forests for image categorization and segmentation. In CVPR. IEEE Computer Society, [5] J. R. R. Uijlings, K. E. A. van de Sande, T. Gevers, and A. W. M. Smeulders. Selective search for object recognition. International Journal of Computer Vision, 104(2): , [6] Jinjun Wang, Jianchao Yang, Kai Yu, Fengjun Lv, Thomas Huang, and Yihong Gong. Locality-constrained linear coding for image classification. In IN: IEEE CONFERENCE ON COMPUTER VISION AND PATTERN CLASSIFICATOIN, [7] Bangpeng Yao, Xiaoye Jiang, Aditya Khosla, Andy Lai Lin, Leonidas J. Guibas, and Li Fei-Fei. Action recognition by learning bases of action attributes and parts. In International Conference on Computer Vision (ICCV), Barcelona, Spain, November [8] Bangpeng Yao, Aditya Khosla, and Li Fei-Fei. Combining randomization and discrimination for fine-grained image categorization. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Springs, USA, June

Aggregating Descriptors with Local Gaussian Metrics

Aggregating Descriptors with Local Gaussian Metrics Aggregating Descriptors with Local Gaussian Metrics Hideki Nakayama Grad. School of Information Science and Technology The University of Tokyo Tokyo, JAPAN nakayama@ci.i.u-tokyo.ac.jp Abstract Recently,

More information

Beyond Bags of Features

Beyond Bags of Features : for Recognizing Natural Scene Categories Matching and Modeling Seminar Instructed by Prof. Haim J. Wolfson School of Computer Science Tel Aviv University December 9 th, 2015

More information

Improving Recognition through Object Sub-categorization

Improving Recognition through Object Sub-categorization Improving Recognition through Object Sub-categorization Al Mansur and Yoshinori Kuno Graduate School of Science and Engineering, Saitama University, 255 Shimo-Okubo, Sakura-ku, Saitama-shi, Saitama 338-8570,

More information

Object Classification Problem

Object Classification Problem HIERARCHICAL OBJECT CATEGORIZATION" Gregory Griffin and Pietro Perona. Learning and Using Taxonomies For Fast Visual Categorization. CVPR 2008 Marcin Marszalek and Cordelia Schmid. Constructing Category

More information

TRANSPARENT OBJECT DETECTION USING REGIONS WITH CONVOLUTIONAL NEURAL NETWORK

TRANSPARENT OBJECT DETECTION USING REGIONS WITH CONVOLUTIONAL NEURAL NETWORK TRANSPARENT OBJECT DETECTION USING REGIONS WITH CONVOLUTIONAL NEURAL NETWORK 1 Po-Jen Lai ( 賴柏任 ), 2 Chiou-Shann Fuh ( 傅楸善 ) 1 Dept. of Electrical Engineering, National Taiwan University, Taiwan 2 Dept.

More information

Ordinal Random Forests for Object Detection

Ordinal Random Forests for Object Detection Ordinal Random Forests for Object Detection Samuel Schulter, Peter M. Roth, Horst Bischof Institute for Computer Graphics and Vision Graz University of Technology, Austria {schulter,pmroth,bischof}@icg.tugraz.at

More information

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun Presented by Tushar Bansal Objective 1. Get bounding box for all objects

More information

A System of Image Matching and 3D Reconstruction

A System of Image Matching and 3D Reconstruction A System of Image Matching and 3D Reconstruction CS231A Project Report 1. Introduction Xianfeng Rui Given thousands of unordered images of photos with a variety of scenes in your gallery, you will find

More information

An Exploration of Computer Vision Techniques for Bird Species Classification

An Exploration of Computer Vision Techniques for Bird Species Classification An Exploration of Computer Vision Techniques for Bird Species Classification Anne L. Alter, Karen M. Wang December 15, 2017 Abstract Bird classification, a fine-grained categorization task, is a complex

More information

The Kinect Sensor. Luís Carriço FCUL 2014/15

The Kinect Sensor. Luís Carriço FCUL 2014/15 Advanced Interaction Techniques The Kinect Sensor Luís Carriço FCUL 2014/15 Sources: MS Kinect for Xbox 360 John C. Tang. Using Kinect to explore NUI, Ms Research, From Stanford CS247 Shotton et al. Real-Time

More information

Beyond bags of features: Adding spatial information. Many slides adapted from Fei-Fei Li, Rob Fergus, and Antonio Torralba

Beyond bags of features: Adding spatial information. Many slides adapted from Fei-Fei Li, Rob Fergus, and Antonio Torralba Beyond bags of features: Adding spatial information Many slides adapted from Fei-Fei Li, Rob Fergus, and Antonio Torralba Adding spatial information Forming vocabularies from pairs of nearby features doublets

More information

The Caltech-UCSD Birds Dataset

The Caltech-UCSD Birds Dataset The Caltech-UCSD Birds-200-2011 Dataset Catherine Wah 1, Steve Branson 1, Peter Welinder 2, Pietro Perona 2, Serge Belongie 1 1 University of California, San Diego 2 California Institute of Technology

More information

A FRAMEWORK OF EXTRACTING MULTI-SCALE FEATURES USING MULTIPLE CONVOLUTIONAL NEURAL NETWORKS. Kuan-Chuan Peng and Tsuhan Chen

A FRAMEWORK OF EXTRACTING MULTI-SCALE FEATURES USING MULTIPLE CONVOLUTIONAL NEURAL NETWORKS. Kuan-Chuan Peng and Tsuhan Chen A FRAMEWORK OF EXTRACTING MULTI-SCALE FEATURES USING MULTIPLE CONVOLUTIONAL NEURAL NETWORKS Kuan-Chuan Peng and Tsuhan Chen School of Electrical and Computer Engineering, Cornell University, Ithaca, NY

More information

Improved Spatial Pyramid Matching for Image Classification

Improved Spatial Pyramid Matching for Image Classification Improved Spatial Pyramid Matching for Image Classification Mohammad Shahiduzzaman, Dengsheng Zhang, and Guojun Lu Gippsland School of IT, Monash University, Australia {Shahid.Zaman,Dengsheng.Zhang,Guojun.Lu}@monash.edu

More information

Preliminary Local Feature Selection by Support Vector Machine for Bag of Features

Preliminary Local Feature Selection by Support Vector Machine for Bag of Features Preliminary Local Feature Selection by Support Vector Machine for Bag of Features Tetsu Matsukawa Koji Suzuki Takio Kurita :University of Tsukuba :National Institute of Advanced Industrial Science and

More information

Video annotation based on adaptive annular spatial partition scheme

Video annotation based on adaptive annular spatial partition scheme Video annotation based on adaptive annular spatial partition scheme Guiguang Ding a), Lu Zhang, and Xiaoxu Li Key Laboratory for Information System Security, Ministry of Education, Tsinghua National Laboratory

More information

Kernel Codebooks for Scene Categorization

Kernel Codebooks for Scene Categorization Kernel Codebooks for Scene Categorization Jan C. van Gemert, Jan-Mark Geusebroek, Cor J. Veenman, and Arnold W.M. Smeulders Intelligent Systems Lab Amsterdam (ISLA), University of Amsterdam, Kruislaan

More information

arxiv: v3 [cs.cv] 3 Oct 2012

arxiv: v3 [cs.cv] 3 Oct 2012 Combined Descriptors in Spatial Pyramid Domain for Image Classification Junlin Hu and Ping Guo arxiv:1210.0386v3 [cs.cv] 3 Oct 2012 Image Processing and Pattern Recognition Laboratory Beijing Normal University,

More information

Deep Tracking: Biologically Inspired Tracking with Deep Convolutional Networks

Deep Tracking: Biologically Inspired Tracking with Deep Convolutional Networks Deep Tracking: Biologically Inspired Tracking with Deep Convolutional Networks Si Chen The George Washington University sichen@gwmail.gwu.edu Meera Hahn Emory University mhahn7@emory.edu Mentor: Afshin

More information

Artistic ideation based on computer vision methods

Artistic ideation based on computer vision methods Journal of Theoretical and Applied Computer Science Vol. 6, No. 2, 2012, pp. 72 78 ISSN 2299-2634 http://www.jtacs.org Artistic ideation based on computer vision methods Ferran Reverter, Pilar Rosado,

More information

Tensor Decomposition of Dense SIFT Descriptors in Object Recognition

Tensor Decomposition of Dense SIFT Descriptors in Object Recognition Tensor Decomposition of Dense SIFT Descriptors in Object Recognition Tan Vo 1 and Dat Tran 1 and Wanli Ma 1 1- Faculty of Education, Science, Technology and Mathematics University of Canberra, Australia

More information

Previously. Part-based and local feature models for generic object recognition. Bag-of-words model 4/20/2011

Previously. Part-based and local feature models for generic object recognition. Bag-of-words model 4/20/2011 Previously Part-based and local feature models for generic object recognition Wed, April 20 UT-Austin Discriminative classifiers Boosting Nearest neighbors Support vector machines Useful for object recognition

More information

III. VERVIEW OF THE METHODS

III. VERVIEW OF THE METHODS An Analytical Study of SIFT and SURF in Image Registration Vivek Kumar Gupta, Kanchan Cecil Department of Electronics & Telecommunication, Jabalpur engineering college, Jabalpur, India comparing the distance

More information

Ranking Error-Correcting Output Codes for Class Retrieval

Ranking Error-Correcting Output Codes for Class Retrieval Ranking Error-Correcting Output Codes for Class Retrieval Mehdi Mirza-Mohammadi, Francesco Ciompi, Sergio Escalera, Oriol Pujol, and Petia Radeva Computer Vision Center, Campus UAB, Edifici O, 08193, Bellaterra,

More information

Part-based and local feature models for generic object recognition

Part-based and local feature models for generic object recognition Part-based and local feature models for generic object recognition May 28 th, 2015 Yong Jae Lee UC Davis Announcements PS2 grades up on SmartSite PS2 stats: Mean: 80.15 Standard Dev: 22.77 Vote on piazza

More information

Random Forest A. Fornaser

Random Forest A. Fornaser Random Forest A. Fornaser alberto.fornaser@unitn.it Sources Lecture 15: decision trees, information theory and random forests, Dr. Richard E. Turner Trees and Random Forests, Adele Cutler, Utah State University

More information

Multimodal Medical Image Retrieval based on Latent Topic Modeling

Multimodal Medical Image Retrieval based on Latent Topic Modeling Multimodal Medical Image Retrieval based on Latent Topic Modeling Mandikal Vikram 15it217.vikram@nitk.edu.in Suhas BS 15it110.suhas@nitk.edu.in Aditya Anantharaman 15it201.aditya.a@nitk.edu.in Sowmya Kamath

More information

Assistive Sports Video Annotation: Modelling and Detecting Complex Events in Sports Video

Assistive Sports Video Annotation: Modelling and Detecting Complex Events in Sports Video : Modelling and Detecting Complex Events in Sports Video Aled Owen 1, David Marshall 1, Kirill Sidorov 1, Yulia Hicks 1, and Rhodri Brown 2 1 Cardiff University, Cardiff, UK 2 Welsh Rugby Union Abstract

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

CS229: Action Recognition in Tennis

CS229: Action Recognition in Tennis CS229: Action Recognition in Tennis Aman Sikka Stanford University Stanford, CA 94305 Rajbir Kataria Stanford University Stanford, CA 94305 asikka@stanford.edu rkataria@stanford.edu 1. Motivation As active

More information

AN IMPROVED K-MEANS CLUSTERING ALGORITHM FOR IMAGE SEGMENTATION

AN IMPROVED K-MEANS CLUSTERING ALGORITHM FOR IMAGE SEGMENTATION AN IMPROVED K-MEANS CLUSTERING ALGORITHM FOR IMAGE SEGMENTATION WILLIAM ROBSON SCHWARTZ University of Maryland, Department of Computer Science College Park, MD, USA, 20742-327, schwartz@cs.umd.edu RICARDO

More information

Developing Open Source code for Pyramidal Histogram Feature Sets

Developing Open Source code for Pyramidal Histogram Feature Sets Developing Open Source code for Pyramidal Histogram Feature Sets BTech Project Report by Subodh Misra subodhm@iitk.ac.in Y648 Guide: Prof. Amitabha Mukerjee Dept of Computer Science and Engineering IIT

More information

Selective Search for Object Recognition

Selective Search for Object Recognition Selective Search for Object Recognition Uijlings et al. Schuyler Smith Overview Introduction Object Recognition Selective Search Similarity Metrics Results Object Recognition Kitten Goal: Problem: Where

More information

Part based models for recognition. Kristen Grauman

Part based models for recognition. Kristen Grauman Part based models for recognition Kristen Grauman UT Austin Limitations of window-based models Not all objects are box-shaped Assuming specific 2d view of object Local components themselves do not necessarily

More information

Recognizing hand-drawn images using shape context

Recognizing hand-drawn images using shape context Recognizing hand-drawn images using shape context Gyozo Gidofalvi Department of Computer Science and Engineering University of California, San Diego La Jolla, CA 92037 gyozo@cs.ucsd.edu Abstract The objective

More information

Combining PGMs and Discriminative Models for Upper Body Pose Detection

Combining PGMs and Discriminative Models for Upper Body Pose Detection Combining PGMs and Discriminative Models for Upper Body Pose Detection Gedas Bertasius May 30, 2014 1 Introduction In this project, I utilized probabilistic graphical models together with discriminative

More information

Spatial Hierarchy of Textons Distributions for Scene Classification

Spatial Hierarchy of Textons Distributions for Scene Classification Spatial Hierarchy of Textons Distributions for Scene Classification S. Battiato 1, G. M. Farinella 1, G. Gallo 1, and D. Ravì 1 Image Processing Laboratory, University of Catania, IT {battiato, gfarinella,

More information

OBJECT CATEGORIZATION

OBJECT CATEGORIZATION OBJECT CATEGORIZATION Ing. Lorenzo Seidenari e-mail: seidenari@dsi.unifi.it Slides: Ing. Lamberto Ballan November 18th, 2009 What is an Object? Merriam-Webster Definition: Something material that may be

More information

Transfer Forest Based on Covariate Shift

Transfer Forest Based on Covariate Shift Transfer Forest Based on Covariate Shift Masamitsu Tsuchiya SECURE, INC. tsuchiya@secureinc.co.jp Yuji Yamauchi, Takayoshi Yamashita, Hironobu Fujiyoshi Chubu University yuu@vision.cs.chubu.ac.jp, {yamashita,

More information

Learning Compact Visual Attributes for Large-scale Image Classification

Learning Compact Visual Attributes for Large-scale Image Classification Learning Compact Visual Attributes for Large-scale Image Classification Yu Su and Frédéric Jurie GREYC CNRS UMR 6072, University of Caen Basse-Normandie, Caen, France {yu.su,frederic.jurie}@unicaen.fr

More information

String distance for automatic image classification

String distance for automatic image classification String distance for automatic image classification Nguyen Hong Thinh*, Le Vu Ha*, Barat Cecile** and Ducottet Christophe** *University of Engineering and Technology, Vietnam National University of HaNoi,

More information

ImageCLEF 2011

ImageCLEF 2011 SZTAKI @ ImageCLEF 2011 Bálint Daróczy joint work with András Benczúr, Róbert Pethes Data Mining and Web Search Group Computer and Automation Research Institute Hungarian Academy of Sciences Training/test

More information

Efficient Kernels for Identifying Unbounded-Order Spatial Features

Efficient Kernels for Identifying Unbounded-Order Spatial Features Efficient Kernels for Identifying Unbounded-Order Spatial Features Yimeng Zhang Carnegie Mellon University yimengz@andrew.cmu.edu Tsuhan Chen Cornell University tsuhan@ece.cornell.edu Abstract Higher order

More information

Generating Object Candidates from RGB-D Images and Point Clouds

Generating Object Candidates from RGB-D Images and Point Clouds Generating Object Candidates from RGB-D Images and Point Clouds Helge Wrede 11.05.2017 1 / 36 Outline Introduction Methods Overview The Data RGB-D Images Point Clouds Microsoft Kinect Generating Object

More information

Semantic-based image analysis with the goal of assisting artistic creation

Semantic-based image analysis with the goal of assisting artistic creation Semantic-based image analysis with the goal of assisting artistic creation Pilar Rosado 1, Ferran Reverter 2, Eva Figueras 1, and Miquel Planas 1 1 Fine Arts Faculty, University of Barcelona, Spain, pilarrosado@ub.edu,

More information

Max-Margin Dictionary Learning for Multiclass Image Categorization

Max-Margin Dictionary Learning for Multiclass Image Categorization Max-Margin Dictionary Learning for Multiclass Image Categorization Xiao-Chen Lian 1, Zhiwei Li 3, Bao-Liang Lu 1,2, and Lei Zhang 3 1 Dept. of Computer Science and Engineering, Shanghai Jiao Tong University,

More information

LETTER Learning Co-occurrence of Local Spatial Strokes for Robust Character Recognition

LETTER Learning Co-occurrence of Local Spatial Strokes for Robust Character Recognition IEICE TRANS. INF. & SYST., VOL.E97 D, NO.7 JULY 2014 1937 LETTER Learning Co-occurrence of Local Spatial Strokes for Robust Character Recognition Song GAO, Student Member, Chunheng WANG a), Member, Baihua

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

IMAGE RETRIEVAL USING VLAD WITH MULTIPLE FEATURES

IMAGE RETRIEVAL USING VLAD WITH MULTIPLE FEATURES IMAGE RETRIEVAL USING VLAD WITH MULTIPLE FEATURES Pin-Syuan Huang, Jing-Yi Tsai, Yu-Fang Wang, and Chun-Yi Tsai Department of Computer Science and Information Engineering, National Taitung University,

More information

arxiv: v1 [cs.lg] 20 Dec 2013

arxiv: v1 [cs.lg] 20 Dec 2013 Unsupervised Feature Learning by Deep Sparse Coding Yunlong He Koray Kavukcuoglu Yun Wang Arthur Szlam Yanjun Qi arxiv:1312.5783v1 [cs.lg] 20 Dec 2013 Abstract In this paper, we propose a new unsupervised

More information

CELLULAR AUTOMATA BAG OF VISUAL WORDS FOR OBJECT RECOGNITION

CELLULAR AUTOMATA BAG OF VISUAL WORDS FOR OBJECT RECOGNITION U.P.B. Sci. Bull., Series C, Vol. 77, Iss. 4, 2015 ISSN 2286-3540 CELLULAR AUTOMATA BAG OF VISUAL WORDS FOR OBJECT RECOGNITION Ionuţ Mironică 1, Bogdan Ionescu 2, Radu Dogaru 3 In this paper we propose

More information

Frequent Inner-Class Approach: A Semi-supervised Learning Technique for One-shot Learning

Frequent Inner-Class Approach: A Semi-supervised Learning Technique for One-shot Learning Frequent Inner-Class Approach: A Semi-supervised Learning Technique for One-shot Learning Izumi Suzuki, Koich Yamada, Muneyuki Unehara Nagaoka University of Technology, 1603-1, Kamitomioka Nagaoka, Niigata

More information

CLASSIFICATION Experiments

CLASSIFICATION Experiments CLASSIFICATION Experiments January 27,2015 CS3710: Visual Recognition Bhavin Modi Bag of features Object Bag of words 1. Extract features 2. Learn visual vocabulary Bag of features: outline 3. Quantize

More information

BUAA AUDR at ImageCLEF 2012 Photo Annotation Task

BUAA AUDR at ImageCLEF 2012 Photo Annotation Task BUAA AUDR at ImageCLEF 2012 Photo Annotation Task Lei Huang, Yang Liu State Key Laboratory of Software Development Enviroment, Beihang University, 100191 Beijing, China huanglei@nlsde.buaa.edu.cn liuyang@nlsde.buaa.edu.cn

More information

Mining Discriminative Adjectives and Prepositions for Natural Scene Recognition

Mining Discriminative Adjectives and Prepositions for Natural Scene Recognition Mining Discriminative Adjectives and Prepositions for Natural Scene Recognition Bangpeng Yao 1, Juan Carlos Niebles 2,3, Li Fei-Fei 1 1 Department of Computer Science, Princeton University, NJ 08540, USA

More information

Deep learning for object detection. Slides from Svetlana Lazebnik and many others

Deep learning for object detection. Slides from Svetlana Lazebnik and many others Deep learning for object detection Slides from Svetlana Lazebnik and many others Recent developments in object detection 80% PASCAL VOC mean0average0precision0(map) 70% 60% 50% 40% 30% 20% 10% Before deep

More information

Effective Classifiers for Detecting Objects

Effective Classifiers for Detecting Objects Effective Classifiers for Detecting Objects Michael Mayo Dept. of Computer Science University of Waikato Private Bag 3105, Hamilton, New Zealand mmayo@cs.waikato.ac.nz Abstract Several state-of-the-art

More information

Image Processing. Image Features

Image Processing. Image Features Image Processing Image Features Preliminaries 2 What are Image Features? Anything. What they are used for? Some statements about image fragments (patches) recognition Search for similar patches matching

More information

Using Geometric Blur for Point Correspondence

Using Geometric Blur for Point Correspondence 1 Using Geometric Blur for Point Correspondence Nisarg Vyas Electrical and Computer Engineering Department, Carnegie Mellon University, Pittsburgh, PA Abstract In computer vision applications, point correspondence

More information

Discrete Optimization of Ray Potentials for Semantic 3D Reconstruction

Discrete Optimization of Ray Potentials for Semantic 3D Reconstruction Discrete Optimization of Ray Potentials for Semantic 3D Reconstruction Marc Pollefeys Joined work with Nikolay Savinov, Christian Haene, Lubor Ladicky 2 Comparison to Volumetric Fusion Higher-order ray

More information

Detecting Thoracic Diseases from Chest X-Ray Images Binit Topiwala, Mariam Alawadi, Hari Prasad { topbinit, malawadi, hprasad

Detecting Thoracic Diseases from Chest X-Ray Images Binit Topiwala, Mariam Alawadi, Hari Prasad { topbinit, malawadi, hprasad CS 229, Fall 2017 1 Detecting Thoracic Diseases from Chest X-Ray Images Binit Topiwala, Mariam Alawadi, Hari Prasad { topbinit, malawadi, hprasad }@stanford.edu Abstract Radiologists have to spend time

More information

A Comparison of SIFT, PCA-SIFT and SURF

A Comparison of SIFT, PCA-SIFT and SURF A Comparison of SIFT, PCA-SIFT and SURF Luo Juan Computer Graphics Lab, Chonbuk National University, Jeonju 561-756, South Korea qiuhehappy@hotmail.com Oubong Gwun Computer Graphics Lab, Chonbuk National

More information

Fast Edge Detection Using Structured Forests

Fast Edge Detection Using Structured Forests Fast Edge Detection Using Structured Forests Piotr Dollár, C. Lawrence Zitnick [1] Zhihao Li (zhihaol@andrew.cmu.edu) Computer Science Department Carnegie Mellon University Table of contents 1. Introduction

More information

CS6670: Computer Vision

CS6670: Computer Vision CS6670: Computer Vision Noah Snavely Lecture 16: Bag-of-words models Object Bag of words Announcements Project 3: Eigenfaces due Wednesday, November 11 at 11:59pm solo project Final project presentations:

More information

CS 231A Computer Vision (Fall 2011) Problem Set 4

CS 231A Computer Vision (Fall 2011) Problem Set 4 CS 231A Computer Vision (Fall 2011) Problem Set 4 Due: Nov. 30 th, 2011 (9:30am) 1 Part-based models for Object Recognition (50 points) One approach to object recognition is to use a deformable part-based

More information

Epithelial rosette detection in microscopic images

Epithelial rosette detection in microscopic images Epithelial rosette detection in microscopic images Kun Liu,3, Sandra Ernst 2,3, Virginie Lecaudey 2,3 and Olaf Ronneberger,3 Department of Computer Science 2 Department of Developmental Biology 3 BIOSS

More information

Patch-based Object Recognition. Basic Idea

Patch-based Object Recognition. Basic Idea Patch-based Object Recognition 1! Basic Idea Determine interest points in image Determine local image properties around interest points Use local image properties for object classification Example: Interest

More information

Part Localization by Exploiting Deep Convolutional Networks

Part Localization by Exploiting Deep Convolutional Networks Part Localization by Exploiting Deep Convolutional Networks Marcel Simon, Erik Rodner, and Joachim Denzler Computer Vision Group, Friedrich Schiller University of Jena, Germany www.inf-cv.uni-jena.de Abstract.

More information

Rushes Video Segmentation Using Semantic Features

Rushes Video Segmentation Using Semantic Features Rushes Video Segmentation Using Semantic Features Athina Pappa, Vasileios Chasanis, and Antonis Ioannidis Department of Computer Science and Engineering, University of Ioannina, GR 45110, Ioannina, Greece

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

Image classification Computer Vision Spring 2018, Lecture 18

Image classification Computer Vision Spring 2018, Lecture 18 Image classification http://www.cs.cmu.edu/~16385/ 16-385 Computer Vision Spring 2018, Lecture 18 Course announcements Homework 5 has been posted and is due on April 6 th. - Dropbox link because course

More information

Comparing Local Feature Descriptors in plsa-based Image Models

Comparing Local Feature Descriptors in plsa-based Image Models Comparing Local Feature Descriptors in plsa-based Image Models Eva Hörster 1,ThomasGreif 1, Rainer Lienhart 1, and Malcolm Slaney 2 1 Multimedia Computing Lab, University of Augsburg, Germany {hoerster,lienhart}@informatik.uni-augsburg.de

More information

K-Means Based Matching Algorithm for Multi-Resolution Feature Descriptors

K-Means Based Matching Algorithm for Multi-Resolution Feature Descriptors K-Means Based Matching Algorithm for Multi-Resolution Feature Descriptors Shao-Tzu Huang, Chen-Chien Hsu, Wei-Yen Wang International Science Index, Electrical and Computer Engineering waset.org/publication/0007607

More information

A Keypoint Descriptor Inspired by Retinal Computation

A Keypoint Descriptor Inspired by Retinal Computation A Keypoint Descriptor Inspired by Retinal Computation Bongsoo Suh, Sungjoon Choi, Han Lee Stanford University {bssuh,sungjoonchoi,hanlee}@stanford.edu Abstract. The main goal of our project is to implement

More information

Learning Visual Similarity Measures for Comparing Never Seen Objects

Learning Visual Similarity Measures for Comparing Never Seen Objects Learning Visual Similarity Measures for Comparing Never Seen Objects Eric Nowak INRIA - INPG - Bertin Technologies 155 r Louis Armand 13290 Aix en Provence - France eric.nowak@inrialpes.fr Frédéric Jurie

More information

Locality-constrained Linear Coding for Image Classification

Locality-constrained Linear Coding for Image Classification Locality-constrained Linear Coding for Image Classification Jinjun Wang, Jianchao Yang,KaiYu, Fengjun Lv, Thomas Huang, and Yihong Gong Akiira Media System, Palo Alto, California Beckman Institute, University

More information

Generic Face Alignment Using an Improved Active Shape Model

Generic Face Alignment Using an Improved Active Shape Model Generic Face Alignment Using an Improved Active Shape Model Liting Wang, Xiaoqing Ding, Chi Fang Electronic Engineering Department, Tsinghua University, Beijing, China {wanglt, dxq, fangchi} @ocrserv.ee.tsinghua.edu.cn

More information

Loose Shape Model for Discriminative Learning of Object Categories

Loose Shape Model for Discriminative Learning of Object Categories Loose Shape Model for Discriminative Learning of Object Categories Margarita Osadchy and Elran Morash Computer Science Department University of Haifa Mount Carmel, Haifa 31905, Israel rita@cs.haifa.ac.il

More information

Automated Canvas Analysis for Painting Conservation. By Brendan Tobin

Automated Canvas Analysis for Painting Conservation. By Brendan Tobin Automated Canvas Analysis for Painting Conservation By Brendan Tobin 1. Motivation Distinctive variations in the spacings between threads in a painting's canvas can be used to show that two sections of

More information

ISyE 6416 Basic Statistical Methods - Spring 2016 Bonus Project: Big Data Analytics Final Report. Team Member Names: Xi Yang, Yi Wen, Xue Zhang

ISyE 6416 Basic Statistical Methods - Spring 2016 Bonus Project: Big Data Analytics Final Report. Team Member Names: Xi Yang, Yi Wen, Xue Zhang ISyE 6416 Basic Statistical Methods - Spring 2016 Bonus Project: Big Data Analytics Final Report Team Member Names: Xi Yang, Yi Wen, Xue Zhang Project Title: Improve Room Utilization Introduction Problem

More information

Analysis: TextonBoost and Semantic Texton Forests. Daniel Munoz Februrary 9, 2009

Analysis: TextonBoost and Semantic Texton Forests. Daniel Munoz Februrary 9, 2009 Analysis: TextonBoost and Semantic Texton Forests Daniel Munoz 16-721 Februrary 9, 2009 Papers [shotton-eccv-06] J. Shotton, J. Winn, C. Rother, A. Criminisi, TextonBoost: Joint Appearance, Shape and Context

More information

Visual Object Recognition

Visual Object Recognition Perceptual and Sensory Augmented Computing Visual Object Recognition Tutorial Visual Object Recognition Bastian Leibe Computer Vision Laboratory ETH Zurich Chicago, 14.07.2008 & Kristen Grauman Department

More information

Learning Visual Semantics: Models, Massive Computation, and Innovative Applications

Learning Visual Semantics: Models, Massive Computation, and Innovative Applications Learning Visual Semantics: Models, Massive Computation, and Innovative Applications Part II: Visual Features and Representations Liangliang Cao, IBM Watson Research Center Evolvement of Visual Features

More information

HIGH spatial resolution Earth Observation (EO) images

HIGH spatial resolution Earth Observation (EO) images JOURNAL OF L A TEX CLASS FILES, VOL. 6, NO., JANUARY 7 A Comparative Study of Bag-of-Words and Bag-of-Topics Models of EO Image Patches Reza Bahmanyar, Shiyong Cui, and Mihai Datcu, Fellow, IEEE Abstract

More information

Bertin Technologies / CNRS LEAR Group INRIA - France

Bertin Technologies / CNRS LEAR Group INRIA - France Learning Visual Distance Function for Identification from one Example. Ei Eric Nowak and dfrederic Jurie Bertin Technologies / CNRS LEAR Group INRIA - France This is an object you've never seen before

More information

A Feature Selection Method to Handle Imbalanced Data in Text Classification

A Feature Selection Method to Handle Imbalanced Data in Text Classification A Feature Selection Method to Handle Imbalanced Data in Text Classification Fengxiang Chang 1*, Jun Guo 1, Weiran Xu 1, Kejun Yao 2 1 School of Information and Communication Engineering Beijing University

More information

Beyond Bags of features Spatial information & Shape models

Beyond Bags of features Spatial information & Shape models Beyond Bags of features Spatial information & Shape models Jana Kosecka Many slides adapted from S. Lazebnik, FeiFei Li, Rob Fergus, and Antonio Torralba Detection, recognition (so far )! Bags of features

More information

Journal of Asian Scientific Research FEATURES COMPOSITION FOR PROFICIENT AND REAL TIME RETRIEVAL IN CBIR SYSTEM. Tohid Sedghi

Journal of Asian Scientific Research FEATURES COMPOSITION FOR PROFICIENT AND REAL TIME RETRIEVAL IN CBIR SYSTEM. Tohid Sedghi Journal of Asian Scientific Research, 013, 3(1):68-74 Journal of Asian Scientific Research journal homepage: http://aessweb.com/journal-detail.php?id=5003 FEATURES COMPOSTON FOR PROFCENT AND REAL TME RETREVAL

More information

Object Category Detection. Slides mostly from Derek Hoiem

Object Category Detection. Slides mostly from Derek Hoiem Object Category Detection Slides mostly from Derek Hoiem Today s class: Object Category Detection Overview of object category detection Statistical template matching with sliding window Part-based Models

More information

Local features and image matching. Prof. Xin Yang HUST

Local features and image matching. Prof. Xin Yang HUST Local features and image matching Prof. Xin Yang HUST Last time RANSAC for robust geometric transformation estimation Translation, Affine, Homography Image warping Given a 2D transformation T and a source

More information

By Suren Manvelyan,

By Suren Manvelyan, By Suren Manvelyan, http://www.surenmanvelyan.com/gallery/7116 By Suren Manvelyan, http://www.surenmanvelyan.com/gallery/7116 By Suren Manvelyan, http://www.surenmanvelyan.com/gallery/7116 By Suren Manvelyan,

More information

A Comparison of l 1 Norm and l 2 Norm Multiple Kernel SVMs in Image and Video Classification

A Comparison of l 1 Norm and l 2 Norm Multiple Kernel SVMs in Image and Video Classification A Comparison of l 1 Norm and l 2 Norm Multiple Kernel SVMs in Image and Video Classification Fei Yan Krystian Mikolajczyk Josef Kittler Muhammad Tahir Centre for Vision, Speech and Signal Processing University

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

Object Recognition. Computer Vision. Slides from Lana Lazebnik, Fei-Fei Li, Rob Fergus, Antonio Torralba, and Jean Ponce

Object Recognition. Computer Vision. Slides from Lana Lazebnik, Fei-Fei Li, Rob Fergus, Antonio Torralba, and Jean Ponce Object Recognition Computer Vision Slides from Lana Lazebnik, Fei-Fei Li, Rob Fergus, Antonio Torralba, and Jean Ponce How many visual object categories are there? Biederman 1987 ANIMALS PLANTS OBJECTS

More information

Real-time Object Detection CS 229 Course Project

Real-time Object Detection CS 229 Course Project Real-time Object Detection CS 229 Course Project Zibo Gong 1, Tianchang He 1, and Ziyi Yang 1 1 Department of Electrical Engineering, Stanford University December 17, 2016 Abstract Objection detection

More information

Fusing shape and appearance information for object category detection

Fusing shape and appearance information for object category detection 1 Fusing shape and appearance information for object category detection Andreas Opelt, Axel Pinz Graz University of Technology, Austria Andrew Zisserman Dept. of Engineering Science, University of Oxford,

More information

Selective Pooling Vector for Fine-grained Recognition

Selective Pooling Vector for Fine-grained Recognition Selective Pooling Vector for Fine-grained Recognition Guang Chen Jianchao Yang Hailin Jin Eli Shechtman Jonathan Brandt Tony X. Han Adobe Research University of Missouri San Jose, CA, USA Columbia, MO,

More information

Pedestrian Detection and Tracking in Images and Videos

Pedestrian Detection and Tracking in Images and Videos Pedestrian Detection and Tracking in Images and Videos Azar Fazel Stanford University azarf@stanford.edu Viet Vo Stanford University vtvo@stanford.edu Abstract The increase in population density and accessibility

More information

ECCV Presented by: Boris Ivanovic and Yolanda Wang CS 331B - November 16, 2016

ECCV Presented by: Boris Ivanovic and Yolanda Wang CS 331B - November 16, 2016 ECCV 2016 Presented by: Boris Ivanovic and Yolanda Wang CS 331B - November 16, 2016 Fundamental Question What is a good vector representation of an object? Something that can be easily predicted from 2D

More information

Data-driven Depth Inference from a Single Still Image

Data-driven Depth Inference from a Single Still Image Data-driven Depth Inference from a Single Still Image Kyunghee Kim Computer Science Department Stanford University kyunghee.kim@stanford.edu Abstract Given an indoor image, how to recover its depth information

More information