Computer Vision: Indexing & Dealing with Outliers
Zur Veranstaltungsseite

RANSAC, Generalized Hough Transform, SIFT, Indexing with Local Features, Visual Vocabularies

Kapitel:

Start Kapitel
00:00:00
Lecture 15: Indexing and Visual Vocabularies
00:00:49
Course Outline
00:02:00
Recap: Local Feature Matching Outline
00:03:34
Recap: Recognition with Local Features
00:04:59
Recap: Fitting an Affine Transformation
00:07:50
Recap: Fitting a Homography
00:15:24
Recap: Object Recognition by Alignment
00:19:57
Topics of This Lecture
00:20:36
Example: Least-Squares Line Fitting
00:21:22
Outliers Affect Least-Squares Fit
00:22:31
Strategy 1: RANSAC [Fischler81]
00:24:56
RANSAC Loop
00:26:05
RANSAC Line Fitting Example
00:30:11
RANSAC: How many samples?
00:34:08
RANSAC: Computed k (p=0.99)
00:38:35
After RANSAC
00:39:55
Example: Finding Feature Matches
00:42:02
Problem with RANSAC
00:43:18
Strategy 2: Generalized Hough Transform
00:47:25
Pose Clustering and Verification with SIFT
00:49:19
Object Recognition Results
00:51:17
Location Recognition
00:52:48
Topics of This Lecture
00:53:53
Application: Mobile Visual Search
00:55:42
Large-Scale Image Matching Problem
00:56:08
Indexing Local Features
00:59:39
Indexing Local Features
01:01:26
Indexing Local Features: Inverted File Index
01:02:44
Text Retrieval vs. Image Search
01:05:21
Visual Words: Main Idea
01:07:55
Indexing with Visual Words
01:08:48
Visual Words
01:12:24
Inverted File for Images of Visual Words
01:15:08
Example: Recognition with Vocabulary Tree
01:16:09
Vocabulary Tree
01:16:25
Vocabulary Tree: Training
01:17:32
Vocabulary Tree: Recognition
01:19:05
Vocabulary Tree: Performance
01:20:29
Vocabulary Size
01:22:02
References and Further Reading