Images and Filters
CSE 576
Ali Farhadi
Many slides from Steve Seitz and Larry Zitnick
AdministraAve Stu
See the setup instrucAons on the course web page
Setup your
Deep Learning
Ali Farhadi
Mohammad Rastegari
CSE 576
Region-based ConvoluNonal
Networks (R-CNNs)
mean Average Precision (mAP)
70%
60%
~1 year
50%
40%
~5 years
30%
Image Sampling
CSE 576
Ali Farhadi
Many slides from Steve Seitz and Larry Zitnick
Image Sampling
F(
)=
F(
)=
Image Scaling
This image is too big to
fit on the screen. How
can we reduce it?
How to gene
Interest points
CSE 576
Ali Farhadi
Many slides from Steve Seitz, Larry Zitnick
How can we find corresponding points?
Not always easy
NASA Mars Rover images
Answer below (look for tiny colored squares
Descriptors III
CSE 576
Ali Farhadi
Many slides from Larry Zitnick, Steve Seitz
How can we find corresponding points?
How can we find correspondences?
SIFT descriptor
Full version
Divide the 16x16 wi
Descriptors II
CSE 576
Ali Farhadi
Many slides from Larry Zitnick, Steve Seitz
How can we find corresponding points?
How can we find correspondences?
SIFT descriptor
Full version
Divide the 16x16 win
Structure From Mo,on
Ali Farhadi
CSE 576
Several slides from Steve Seitz, Rick Szeliski, Mar,al Hebert, and Noha Snavely
Structure from mo,on
aka bundle adjustment
Edge Detection
CSE 576
Ali Farhadi
Many slides from Steve Seitz and Larry Zitnick
Edge
Attneave's Cat (1954)
Origin of edges
surface normal discontinuity
depth discontinuity
surface color discontinuit
Reconstruction
CSE 576
Ali Farhadi
Several slides from Steve Seitz, Larry Zitnick, Lana Lazebnik, Carlos Hernndez, George Vogiatzis,Yasutaka Furukawa
3d model
Digital copy of real object
Allows u
Stereo
CSE 576
Ali Farhadi
Several slides from Larry Zitnick and Steve Seitz
Why do we perceive depth?
What do humans use as depth cues?
Motion
Convergence
When watchin
Stereo II
CSE 576
Ali Farhadi
Several slides from Larry Zitnick and Steve Seitz
Camera parameters
A camera is described by several parameters
Translation T of the optical ce
Images and Filters
CSE 576
Ali Farhadi
Many slides from Steve Seitz and Larry Zitnick
Administrative Stuff
See the setup instructions on the course web page
Setup your environment
Project
Topic
T
Computer Vision
CSE 576
Ali Farhadi
Many slides from Steve Seitz, Larry Zitnick, Yang Wang
Course Information
Time:
Monday, Wednesday 1:30-2:50
Location:
MGH 238
Contact:
[email protected] , CSE 652
Recognition
Part I
CSE 576
What we have seen so far:
Vision as Measurement Device
Real-time stereo on Mars
Physics-based Vision
Structure from Motion
Virtualized Reality
Slide Credit: Alyosha
Visual R
Object Detec)on
Ali Farhadi
Mohammad Rastegari
CSE 576
Object Recogni)on
Dog
Person
Chair
Object Detec)on
Person
Dog Dog
Sliding Window
Sliding Window
Image Categor
Geometric Transformations
CSE 576
Ali Farhadi
Many slides from Steve Seitz and Larry Zitnick
What are geometric transformations?
Translation
Preserves: Orientation
Translation and rotation
Scale
Simil
Computer Vision
CSE 576
Ali Farhadi
Many slides from Steve Seitz, Larry Zitnick, Yang Wang
Course InformaGon
Time:
Monday, Wednesday 1:30-2:50
LocaGon:
MGH
Edge Detec)on
CSE 576
Ali Farhadi
Many slides from Steve Seitz and Larry Zitnick
Edge
ABneave's Cat (1954)
Origin of edges
surface normal discontinuity
depth discontinuity
su
Geometric Transformations
CSE 576
Ali Farhadi
Many slides from Steve Seitz and Larry Zitnick
What are geometric transformations?
Translation
Preserves: OrientaBon
Translation and r
Interest points
CSE 576
Ali Farhadi
Many slides from Steve Seitz, Larry Zitnick
How can we find corresponding points?
Not always easy
NASA Mars Rover images
Answer below (loo
Descriptors
CSE 576
Ali Farhadi
Many slides from Larry Zitnick, Steve Seitz
How can we find corresponding points?
How can we find correspondences?
How do we describe an image pa
Face Recogni+on
CSE 576
Face recogni+on: once youve
detected and cropped a face, try to
recognize it
Detection
Recognition
Sally
Face recogni+on: overview
Typical scenar
Final Presenta,on
Logis,cs
Friday, June 3rd, 10am
Start preparing at 9:30
CSE atrium
Pizza will be there at 11:30
To print your poster
Talk to Hessam
Outlets
Face Detec(on
CSE 576
Face detec(on
State-of-the-art face detec(on demo
(Courtesy Boris Babenko)
Face detec(on and recogni(on
Detec(on
Recogni(on
Sally
Face detec(on
W
Deep Object Detec*on
Ali Farhadi
Mohammad Rastegari
CSE 576
Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. Deep Residual Learning for Image Recognition. arXiv 2015.
So Far
Ba
Mo#on and Op#cal Flow
Ali Farhadi
CSE 576
Several slides from Ce Liu, Steve Seitz, Larry Zitnick
We live in a moving world
Perceiving, understanding and pre
Object Detec)on
Ali Farhadi
Mohammad Rastegari
CSE 576
So Far
Support Vector Machines (SVM)
Pedestrian Detec)on by HOG
Implicit Shape Models
Detector Evalua)on
PASC
Image Stitching
Ali Farhadi
CSE 576
Several slides from Rick Szeliski, Steve Seitz, Derek Hoiem, and Ira Kemelmacher
Combine two or more overlapping images
to make one larger image
Add example
Slide