A Codebook-Free and Annotation-free Approach for Fine-Grained Image Categorization Authors Bangpeng Yao et al. Presenter Hyung-seok Lee ( 이형석 ) CVPR 2012.

Slides:

Advertisements

Similar presentations

Negative Selection Algorithms at GECCO /22/2005.

Advertisements

Location Recognition Given: A query image A database of images with known locations Two types of approaches: Direct matching: directly match image features.

Relevant characteristics extraction from semantically unstructured data PhD title : Data mining in unstructured data Daniel I. MORARIU, MSc PhD Supervisor:

Zhimin CaoThe Chinese University of Hong Kong Qi YinITCS, Tsinghua University Xiaoou TangShenzhen Institutes of Advanced Technology Chinese Academy of.

Human Identity Recognition in Aerial Images Omar Oreifej Ramin Mehran Mubarak Shah CVPR 2010, June Computer Vision Lab of UCF.

Bag-of-features models. Origin 1: Texture recognition Texture is characterized by the repetition of basic elements or textons For stochastic textures,

A Nonparametric Treatment for Location/Segmentation Based Visual Tracking Le Lu Integrated Data Systems Dept. Siemens Corporate Research, Inc. Greg Hager.

Large-Scale Object Recognition with Weak Supervision

Ghunhui Gu, Joseph J. Lim, Pablo Arbeláez, Jitendra Malik University of California at Berkeley Berkeley, CA

Bag-of-features models Many slides adapted from Fei-Fei Li, Rob Fergus, and Antonio Torralba.

Self Taught Learning : Transfer learning from unlabeled data Presented by: Shankar B S DMML Lab Rajat Raina et al, CS, Stanford ICML 2007.

Recognition using Regions CVPR Outline Introduction Overview of the Approach Experimental Results Conclusion.

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

Lecture 28: Bag-of-words models

Support Vector Machines Pattern Recognition Sergios Theodoridis Konstantinos Koutroumbas Second Edition A Tutorial on Support Vector Machines for Pattern.

Bag-of-features models

Local Features and Kernels for Classification of Object Categories J. Zhang --- QMUL UK (INRIA till July 2005) with M. Marszalek and C. Schmid --- INRIA.

5/30/2006EE 148, Spring Visual Categorization with Bags of Keypoints Gabriella Csurka Christopher R. Dance Lixin Fan Jutta Willamowski Cedric Bray.

K-means Based Unsupervised Feature Learning for Image Recognition Ling Zheng.

Spatial Pyramid Pooling in Deep Convolutional

(Fri) Young Ki Baik Computer Vision Lab.

Generic object detection with deformable part-based models

Review: Intro to recognition Recognition tasks Machine learning approach: training, testing, generalization Example classifiers Nearest neighbor Linear.

Bag-of-features models. Origin 1: Texture recognition Texture is characterized by the repetition of basic elements or textons For stochastic textures,

Multiclass object recognition

Thien Anh Dinh1, Tomi Silander1, Bolan Su1, Tianxia Gong

Learning Visual Bits with Direct Feature Selection Joel Jurik 1 and Rahul Sukthankar 2,3 1 University of Central Florida 2 Intel Research Pittsburgh 3.

Object Bank Presenter ： Liu Changyu Advisor ： Prof. Alex Hauptmann Interest ： Multimedia Analysis April 4 th, 2013.

PageRank for Product Image Search Kevin Jing (Googlc IncGVU, College of Computing, Georgia Institute of Technology) Shumeet Baluja (Google Inc.) WWW 2008.

Nonparametric Part Transfer for Fine-grained Recognition Presenter Byungju Kim.

“Secret” of Object Detection Zheng Wu (Summer intern in MSRNE) Sep. 3, 2010 Joint work with Ce Liu (MSRNE) William T. Freeman (MIT) Adam Kalai (MSRNE)

Why Categorize in Computer Vision ?. Why Use Categories? People love categories!

1 Action Classification: An Integration of Randomization and Discrimination in A Dense Feature Representation Computer Science Department, Stanford University.

Bag-of-features models. Origin 1: Texture recognition Texture is characterized by the repetition of basic elements or textons For stochastic textures,

Learning Collections of Parts for Object Recognition and Transfer Learning University of Illinois at Urbana- Champaign.

Beyond Sliding Windows: Object Localization by Efficient Subwindow Search The best paper prize at CVPR 2008.

Efficient Subwindow Search: A Branch and Bound Framework for Object Localization ‘PAMI09 Beyond Sliding Windows: Object Localization by Efficient Subwindow.

BAGGING ALGORITHM, ONLINE BOOSTING AND VISION Se – Hoon Park.

 Detecting system  Training system Human Emotions Estimation by Adaboost based on Jinhui Chen, Tetsuya Takiguchi, Yasuo Ariki （ Kobe University ） User's.

Distributed Representative Reading Group. Research Highlights 1Support vector machines can robustly decode semantic information from EEG and MEG 2Multivariate.

Date : 2013/03/18 Author : Jeffrey Pound, Alexander K. Hudek, Ihab F. Ilyas, Grant Weddell Source : CIKM’12 Speaker : Er-Gang Liu Advisor : Prof. Jia-Ling.

Epitomic Location Recognition A generative approach for location recognition K. Ni, A. Kannan, A. Criminisi and J. Winn In proc. CVPR Anchorage,

Sparse Bayesian Learning for Efficient Visual Tracking O. Williams, A. Blake & R. Cipolloa PAMI, Aug Presented by Yuting Qi Machine Learning Reading.

Gang WangDerek HoiemDavid Forsyth. INTRODUCTION APROACH (implement detail) EXPERIMENTS CONCLUSION.

Hierarchical Matching with Side Information for Image Classification

Learning Features and Parts for Fine-Grained Recognition Authors: Jonathan Krause, Timnit Gebru, Jia Deng, Li-Jia Li, Li Fei-Fei ICPR, 2014 Presented by:

Poselets: Body Part Detectors Trained Using 3D Human Pose Annotations ZUO ZHEN 27 SEP 2011.

Object Recognition as Ranking Holistic Figure-Ground Hypotheses Fuxin Li and Joao Carreira and Cristian Sminchisescu 1.

Convolutional Restricted Boltzmann Machines for Feature Learning Mohammad Norouzi Advisor: Dr. Greg Mori Simon Fraser University 27 Nov

Goggle Gist on the Google Phone A Content-based image retrieval system for the Google phone Manu Viswanathan Chin-Kai Chang Ji Hyun Moon.

Fine-grained Fine-grained Recognition( 细粒度分类 ) 沈志强.

Finding Clusters within a Class to Improve Classification Accuracy Literature Survey Yong Jae Lee 3/6/08.

PANDA: Pose Aligned Networks for Deep Attribute Modeling Ning Zhang 1,2 Manohar Paluri 1 Marć Aurelio Ranzato 1 Trevor Darrell 2 Lumbomir Boudev 1 1 Facebook.

Week 3 Emily Hand UNR. Online Multiple Instance Learning The goal of MIL is to classify unseen bags, instances, by using the labeled bags as training.

1 Bilinear Classifiers for Visual Recognition Computational Vision Lab. University of California Irvine To be presented in NIPS 2009 Hamed Pirsiavash Deva.

Hybrid Deep Learning for Reflectance Confocal Microscopy Skin Images

Compact Bilinear Pooling

Object detection with deformable part-based models

Learning Mid-Level Features For Recognition

Table 1. Advantages and Disadvantages of Traditional DM/ML Methods

Mixtures of Gaussians and Advanced Feature Encoding

Group Norm for Learning Latent Structural SVMs

CS 1674: Intro to Computer Vision Scene Recognition

Principal Component Analysis

RCNN, Fast-RCNN, Faster-RCNN

Human-object interaction

Motivation It can effectively mine multi-modal knowledge with structured textural and visual relationships from web automatically. We propose BC-DNN method.

Visual Grounding.

Do Better ImageNet Models Transfer Better?

Presentation transcript:

A Codebook-Free and Annotation-free Approach for Fine-Grained Image Categorization Authors Bangpeng Yao et al. Presenter Hyung-seok Lee ( 이형석 ) CVPR 2012

What is Fine-Grained Categorization? Task of classifying object that belong to the same basic-level category Bird species Flowers Stonefly larvae 2 Red Eyed Vireo Red Headed Woodpecker

Related Work (1) Codebook-based approach Encoding local image patches to visual codewords Large loss of finer details important for Fine-grained 3

Related Work(2) Annotation-based approach Human annotation of object attributes or keypoint Labor cost high and Non-automatic 4 Ref : Multiclass Recognition and Part Localization with Humans in the Loop, Catherine Wah et al.

Overview 5 Template Matching Feature response map Novel bagging-based algorithm 1. Feature Extraction 2. Feature Representation 3. Classification

Template Matching Generate a large number of templates by randomly sampling rectangular regions from all training images 6

Input image is represented by the response score of matching it self with each of the template It can captures the subtle distinctions 7

Feature representation 8 1 step : Three-largest response score 2 step : Largest score on each region

9 Final image representation is formed by concatenating the pooling results of all the templates on all image scales No. templates x No. scales X 7 dimensional

Bagging Based Classification 10 Motivation Large # of template  high dimensional feature It provide richer image representation & capture more subtle visual distinction

11 But over-complete & non-discriminative element Conventional classification such as single SVM suffer from overfitting

12 Traditional bagging method Randomly select element from image feature vector Aggregating set of classifiers Ref : Model averaging Avoid overfitting

13 Novel bagging-based algorithm Full usage of available template matching results

14 Guarantees that the correlation between the classifiers are small C : Regularization T : Correlation tolerance

Experiments CUB-2010 (200 bird species, bounding box) 14 birds species from the vireos and woodpeckers families Experiment setup Randomly generate 100 template per training image Scaling factor : x 100 x 3 x 7 = dimension for each image Total of 420 training images and 492 test images Bagging repetition number : 80 15

Result Compare with state-of-the-art 16

Our-SVM w.r.t # templates 17 Analysis of bagging-based classification

Robustness to non-accurate object locations 18

Conclusion A codebook-free and annotation-free fine-grained By image template matching Bagging-based method Deal with redundant and noisy large-dimensional features. 19