Explore the second lecture in a comprehensive computer vision course covering advanced topics like object recognition, facial analysis, action detection, and video surveillance. Delve into cutting-edge techniques for image and video processing, including semantic segmentation, high-density crowd analysis, and multi-object tracking. Learn about the evolution of computer vision capabilities compared to human vision, from historical approaches like template matching to modern deep learning methods. Gain insights into practical applications such as biometrics, visual business recognition, and vision-based interaction for games. Examine case studies in aerial video detection and police chase scenarios to understand real-world implementations of computer vision algorithms.
Overview
Syllabus
Intro
Object Recognition
Object localization
Human Detection
Semantic Segmentation: Results
Semantic part labeling
Face Recognition
Open Universe Face Identification
Facial expression
Fatigue detection
Lip-reading
High Density Crowded Scenes
Counting
Visual Business Recognition
Biometrics
Smile detection
Vision-based interaction (and games)
Video Clip
Sequences of Images
Action recognition - UCF101
Action detection
Video segmentation
Cross-view action synthesis
Detection in aerial videos
(Object) Tracking
Tracking (multi-object)
Video Surveillance and Monitoring
Naïve approach: Template Matching
"Bag-of-Words" Representation
"Bag-of-Words" (BOW) Histograms
BOW Representation
Computer vs. Human Vision?
50 Years Ago
5 years ago
Police chase
Taught by
UCF CRCV
