pedestrian recognition

21
Pedestrian Recognition Machine Perception and Modeling of Human Behavior Manfred Lau

Upload: benjy

Post on 13-Jan-2016

53 views

Category:

Documents


1 download

DESCRIPTION

Pedestrian Recognition. Machine Perception and Modeling of Human Behavior Manfred Lau. Pedestrian Recognition. Oren, Papageorgiou, Sinha, Osuna, Poggio. Pedestrian Detection Using Wavelet Templates. CVPR 1997. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Pedestrian Recognition

Pedestrian Recognition

Machine Perception and Modeling of Human Behavior

Manfred Lau

Page 2: Pedestrian Recognition

Pedestrian Recognition

Oren, Papageorgiou, Sinha, Osuna, Poggio. Pedestrian Detection Using Wavelet Templates. CVPR 1997.

Papageorgiou, and Poggio. Trainable Pedestrian Detection. International Conference on Image Processing 1999.

Page 3: Pedestrian Recognition

Motivation

Recognition system inside vehicles

Valerie – detect and greet those who stop in front of the booth

Page 4: Pedestrian Recognition

Overview

Positive samples Negative samples

Classifier

Page 5: Pedestrian Recognition

Wavelet Template

-1 1

vertical wavelet

Average of many samples

Compute coefficient for each RGB channel and take largest absolute value

Vertical wavelet identifies “vertical color differences”

Page 6: Pedestrian Recognition

Wavelet Template

-1 1

vertical horizontal diagonal

-1

1

-11

Average of many samples

Page 7: Pedestrian Recognition

Features

Each image is one instance with 1326 features and one classification

Same thing for negative samples

Page 8: Pedestrian Recognition

Test case 282 positive samples, 236 negative

samples for training 20 positives and 20 negatives for testing

Some Positive Samples

Page 9: Pedestrian Recognition

Some negative samples

Page 10: Pedestrian Recognition

Results

Nearest neighbor classifier 95% accuracy

Decision tree classifier 90% accuracy

2 false positives 3 false positives, 1 false negative

Page 11: Pedestrian Recognition

10-fold cross validation Test case: 302 positives, 256 negatives

Nearest neighbor 94.27% 30 false positives, 2 false negatives

Decision tree 86.74% 47 false positives, 27 false negatives

Page 12: Pedestrian Recognition

Incremental bootstrapping Use nearest neighbor

But problem with many false positives

Page 13: Pedestrian Recognition

Incremental bootstrapping Took database of 558 total samples After bootstrapping, 656 total samples

Page 14: Pedestrian Recognition

Bootstrapping

Page 15: Pedestrian Recognition

Result A completely new

test image

Before bootstrapping 85.06% accurate, 65 false pos, 0 false neg

After bootstrapping 90.11% accurate, 43 false pos, 0 false neg

Page 16: Pedestrian Recognition

Result Another new

test image

Before bootstrapping 75.86% accurate, 100 false pos, 5 false neg

After bootstrapping 81.15% accurate, 77 false pos, 5 false neg

Page 17: Pedestrian Recognition

Splitted up into 560 images, about 30 classified as positive

Some false positives

true positives

Page 18: Pedestrian Recognition

Results

Page 19: Pedestrian Recognition

Less features

Take average coefficients across many positive samples

Pick those features that are darkest/lightest can use much less than 1326 features, for faster classification

Page 20: Pedestrian Recognition

Conclusions

Can detect positive samples well, but many false positives

Bootstrapping on more and more new images will decrease false positives (I’m not doing enough of this)

Page 21: Pedestrian Recognition

Limitations Recognize only template,

other objects may be similar

Difficult to define what is a negative sample

What if pedestrians are partially occluded?