Computer Vision: Sliding-Window based Object Detection II (Di, 02.12.2014)

Anmeldung erforderlich

RWTH

Für RWTH-Angehörige und aus dem RWTH-Netz verfügbar

Anmelden
  • Einbetten

Beschreibung:

Boosting, AdaBoost, Viola-Jones Detector

Kapitel:

00:00:00
Lecture 11: Sliding-Window based Object Detection II
00:00:24
Course Outline
00:01:12
Topics of This Lecture
00:01:50
Recap: Sliding-Window Object Detection
00:04:40
Recap: Support Vector Machines (SVM)
00:13:56
Recap: Non-Linear SVMs
00:16:56
Recap: Gradient-based Representations
00:18:41
Recap: HOG Descriptor Processing Chain
00:21:33
Recap: Pedestrian Detection with HOG
00:22:27
Recap: Non-Maximum Suppression
00:23:13
Applications: Mobile Robot Navigation
00:31:24
Classifier Construction: Many Choices...
00:31:40
Boosting
00:33:48
AdaBoost: Intuition
00:36:31
AdaBoost - Formalization
00:38:51
AdaBoost: Detailed Training Algorithm
00:44:00
AdaBoost: Recognition
00:47:08
Example: Face Detection
00:48:49
Feature extraction
00:54:29
Example
00:55:36
Large Library of Filters
00:58:34
AdaBoost for Feature+Classifier Selection
01:03:46
AdaBoost for Efficient Feature Selection
01:05:24
Cascading Classifiers for Detection
01:09:28
Cascading Classifiers
01:11:04
Viola-Jones Face Detector: Summary
01:13:11
Cascading Classifiers
01:14:45
Viola-Jones Face Detector: Summary
01:18:02
Practical Issue: Bootstrapping
01:20:43
Viola-Jones Face Detector: Results
01:22:21
You Can Try It At Home...
01:23:05
Example Application
01:25:01
Summary: Sliding-Windows
01:26:44
Feature Computation Trade-Off
01:28:22
What Slows Down HOG (CUDA Implem.)
01:29:30
References and Further Reading