A Bayesian Framework for Unsupervised One-Shot Learning of Object Categories



Similar documents
Object class recognition using unsupervised scale-invariant learning

Object Class Recognition by Unsupervised Scale-Invariant Learning

3D Model based Object Class Detection in An Arbitrary View

Semantic Image Segmentation and Web-Supervised Visual Learning

Principled Hybrids of Generative and Discriminative Models

Object Recognition. Selim Aksoy. Bilkent University

Discovering objects and their location in images

Local features and matching. Image classification & object localization

Basics of Statistical Machine Learning

Solving Big Data Problems in Computer Vision with MATLAB Loren Shure

Finding people in repeated shots of the same scene

Course: Model, Learning, and Inference: Lecture 5

STA 4273H: Statistical Machine Learning

Statistical Machine Learning from Data

Scale-Invariant Object Categorization using a Scale-Adaptive Mean-Shift Search

Cees Snoek. Machine. Humans. Multimedia Archives. Euvision Technologies The Netherlands. University of Amsterdam The Netherlands. Tree.

Probabilistic Latent Semantic Analysis (plsa)

Monotonicity Hints. Abstract

Statistics Graduate Courses

How To Register Point Sets

1 Prior Probability and Posterior Probability

Discovering objects and their location in images

Learning Spatial Context: Using Stuff to Find Things

Multi-Class Active Learning for Image Classification

Practical Tour of Visual tracking. David Fleet and Allan Jepson January, 2006

Compacting ConvNets for end to end Learning

Lecture 9: Introduction to Pattern Analysis

Visual Categorization with Bags of Keypoints

Tracking in flussi video 3D. Ing. Samuele Salti

Convolutional Feature Maps

Probabilistic Models for Big Data. Alex Davies and Roger Frigola University of Cambridge 13th February 2014

CS 2750 Machine Learning. Lecture 1. Machine Learning. CS 2750 Machine Learning.

3F3: Signal and Pattern Processing

An Analysis of Single-Layer Networks in Unsupervised Feature Learning

Learning Motion Categories using both Semantic and Structural Information

Bayesian Image Super-Resolution

Transfer Learning by Borrowing Examples for Multiclass Object Detection

Are we ready for Autonomous Driving? The KITTI Vision Benchmark Suite

Semantic Recognition: Object Detection and Scene Segmentation

High Level Describable Attributes for Predicting Aesthetics and Interestingness

Package EstCRM. July 13, 2015

Bayesian Statistics in One Hour. Patrick Lam

Bayesian networks - Time-series models - Apache Spark & Scala

Supporting Online Material for

UNSUPERVISED COSEGMENTATION BASED ON SUPERPIXEL MATCHING AND FASTGRABCUT. Hongkai Yu and Xiaojun Qi

Object Categorization using Co-Occurrence, Location and Appearance

Introduction to Deep Learning Variational Inference, Mean Field Theory

Image and Video Understanding

Automatic Attribute Discovery and Characterization from Noisy Web Data

MS1b Statistical Data Mining

HT2015: SC4 Statistical Data Mining and Machine Learning

Unsupervised Joint Alignment of Complex Images

MA2823: Foundations of Machine Learning

Vehicle Tracking by Simultaneous Detection and Viewpoint Estimation

Now we begin our discussion of exploratory data analysis.

Introduction to Data Mining

Fast R-CNN Object detection with Caffe

Forecasting Trade Direction and Size of Future Contracts Using Deep Belief Network

Inference of Probability Distributions for Trust and Security applications

Using SAS PROC MCMC to Estimate and Evaluate Item Response Theory Models

Object Class Recognition using Images of Abstract Regions

Markov Chain Monte Carlo Simulation Made Simple

11. Time series and dynamic linear models

Unsupervised Learning of Invariant Feature Hierarchies with Applications to Object Recognition

Università degli Studi di Bologna

Part 2: One-parameter models

Predict Influencers in the Social Network

Detection. Perspective. Network Anomaly. Bhattacharyya. Jugal. A Machine Learning »C) Dhruba Kumar. Kumar KaKta. CRC Press J Taylor & Francis Croup

Machine Learning for Data Science (CS4786) Lecture 1

Gaussian Processes to Speed up Hamiltonian Monte Carlo

Machine Learning and Data Analysis overview. Department of Cybernetics, Czech Technical University in Prague.

Section 5. Stan for Big Data. Bob Carpenter. Columbia University

MACHINE LEARNING IN HIGH ENERGY PHYSICS

Transform-based Domain Adaptation for Big Data

Whose Vote Should Count More: Optimal Integration of Labels from Labelers of Unknown Expertise

Multi-modal Human-Computer Interaction. Attila Fazekas.

Manifold Learning with Variational Auto-encoder for Medical Image Analysis

FastKeypointRecognitioninTenLinesofCode

Efficient Cluster Detection and Network Marketing in a Nautural Environment

Fast Matching of Binary Features

Image Segmentation and Registration

A Hybrid Neural Network-Latent Topic Model

Transcription:

A Bayesian Framework for Unsupervised One-Shot Learning of Object Categories Li Fei-Fei 1 Rob Fergus 1,2 Pietro Perona 1 1. California Institute of Technology 2. Oxford University

Single Object Object Category Lowe et al. Schmidet al.

Algorithm Training Examples Categories Rowley et al. ~500 Faces Schneiderman, et al. ~2,000 Faces, Cars Viola et al. ~10,000 Faces Burl, et al. Weber, et al. Fergus, et al. 200 ~ 400 Faces, Motorbikes, Spotted cats, Airplanes, Cars

Algorithm Training Examples Categories Rowley et al. ~500 Faces Schneiderman, et al. ~2,000 Faces, Cars Viola et al. ~10,000 Faces Burl, et al. Weber, et al. Fergus, et al. 200 ~ 400 Faces, Motorbikes, Spotted cats, Airplanes, Cars

Words of wisdom from statisticians error Bayes error Test ~5 x (# of params) N* Train # training examples (N)

Goal One-Shot learning of object categories.

There are about 30,000 visual categories. Learn 4.5 categories per day 18 years! At age 6, a child can learn roughly all 30,000 visual categories. (13.5/day) P. Buegel, 1559

S. Savarese, 2003

P. Buegel, 1562

Sneak preview Algorithm Training Examples Categories Results (error) Burl, et al. Weber, et al. Fergus, et al. 200 ~ 400 Faces, Motorbikes, Spotted cats, Airplanes, Cars 5.6-10 % Bayesian One-Shot 1 ~ 5 Faces, Motorbikes, Spotted cats, Airplanes 8 15 %

Outline of main points Intuition: Use prior knowledge Approach: Full Bayesian Implementation: Probabilistic model Variational EM Experiments

Prior knowledge about objects I Appearance Shape unlikely likely

Prior knowledge about objects II Appearance Shape unlikely likely

Bayesian framework P(object test, train) vs. P(clutter test, train) Bayes Rule p( test object, train) p(object) Expansion by parametrization p(test θ,object) p( θ object, train) dθ

Bayesian framework P(object test, train) vs. P(clutter test, train) Bayes Rule p( test object, train) p(object) Expansion by parametrization p(test θ,object) p( θ object, train) dθ Previous Work: ( ML ) δ θ θ

Model (θ) space ML θ ML p ( θ 1 ) δ (.) p ( θ ML n ) δ (.) ML p ( θ 2 ) δ (.) θ 1 θ n θ 2

Bayesian framework P(object test, train) vs. P(clutter test, train) Bayes Rule p( test object, train) p(object) Expansion by parametrization p(test θ,object) p( θ object, train) dθ

Bayesian framework P(object test, train) vs. P(clutter test, train) Bayes Rule p( test object, train) p(object) Expansion by parametrization p(test θ,object) p( θ object, train) dθ One-Shot learning: p ( train θ,object) p( θ )

Model (θ) space θ 1 θ 2 θ n

Representing θ Main issues: measuring the similarity of parts representing the configuration of parts φ Fischler & Elschlager 1973 φ Yuille 91 φ Brunelli & Poggio 93 φ Lades, v.d. Malsburg et al. 93 φ Cootes, Lanitis, Taylor et al. 95 φ Amit & Geman 95, 99 φ Perona et al. 95, 96, 98, 00, 03 φ Agarwal & Roth 02

model (θ) space Model Structure Each object model θ Gaussian shape pdf Gaussian part appearance pdf θ 1 θ n θ 2

model (θ) space Model Structure Each object model θ Gaussian shape pdf Gaussian part appearance pdf θ 1 θ n θ 2 model distribution: p(θ) conjugate priors

Variational EM Random initialization new θ s E-Step M-Step new estimate of p(θ train) Attias, Hinton, Minka, etc. prior knowledge of p(θ)

Maximum Likelihood method Bayesian Learning method

Summary of main points Goal: one-shot learning Intuition: use general prior knowledge Approach: full Bayesian Marginalize over all θ Implementation: Probability models Variational EM

Experiments Training: Testing: 1-6 images 50 fg/ 50 bg images (randomly drawn) object present/absent Datasets faces airplanes spotted cats motorbikes [www.vision.caltech.edu]

Faces Motorbikes Airplanes Spotted cats

No Manual Preprocessing No labeling No segmentation No alignment

Experiments: obtaining priors airplanes Prior distr. Learning spotted cats motorbikes Miller, et al. 00

One more word about priors Data Prior distr. Expertise Non-informative Others

Number of training examples

Number of training examples

Number of training examples

Number of training examples

Algorithm Training Examples Categories Results (error) Burl, et al. Weber, et al. Fergus, et al. 200 ~ 400 Faces, Motorbikes, Spotted cats, Airplanes, Cars 5.6-10 % Bayesian One-Shot 1 ~ 5 Faces, Motorbikes, Spotted cats, Airplanes 8 15 %

Summary Learning categories w/ one example is possible Decreased # of training example from ~300 to 1~5 Bayesian treatment Priors from unrelated categories are useful Future Work Incremental learning More categories (e.g. >100) More on priors

Acknowledgements David MacKay, Cambridge University Brian Ripley, Oxford University Yaser Abu-Mostafa, Caltech Andrew Zisserman, Oxford University Leung, Burl, Perona, ICCV 95 Burl, Leung, Perona, CVPR 98 Weber, Welling, Perona, ECCV 00 Weber, Welling, Perona, CVPR 00 Set up constellation model Affine invariance in recognition Unsupervised learning, two-stage learning of shape & appearance Fergus, Perona, Zisserman, CVPR 03 Fei-Fei, Fergus, Perona, ICCV 03 Simultaneous learning of shape & appearance, scale invariance Full Bayesian treatment, One-Shot learning