Accurate Video Localization

Slides:



Advertisements
Similar presentations
Mining User Similarity Based on Location History Yu Zheng, Quannan Li, Xing Xie Microsoft Research Asia.
Advertisements

Computing a tree Genome 559: Introduction to Statistical and Computational Genomics Prof. James H. Thomas.
Data Compression CS 147 Minh Nguyen.
Spatial-Temporal Consistency in Video Disparity Estimation ICASSP 2011 Ramsin Khoshabeh, Stanley H. Chan, Truong Q. Nguyen.
$100 $200 $300 $400 $500 $100 $200 $300 $400 $500 $100 $200 $300 $400 $500 $100 $200 $300 $400 $500 $100 $200 $300 $400 $500 $100 $200 $300.
Multiple Frame Motion Inference Using Belief Propagation Jiang Gao Jianbo Shi Presented By: Gilad Kapelushnik Visual Recognition, Spring 2005, Technion.
Computer Science 320 Parallel Computing Design Patterns.
Computing a tree Genome 559: Introduction to Statistical and Computational Genomics Prof. James H. Thomas.
Brandon Andrews CS6030.  What is a phylogenetic tree?  Goals in a phylogenetic tree generator  Distance based method  Fitch-Margoliash Method Example.
Christoph F. Eick Questions and Topics Review Dec. 1, Give an example of a problem that might benefit from feature creation 2.Compute the Silhouette.
A Bayesian algorithm for tracking multiple moving objects in outdoor surveillance video Department of Electrical Engineering and Computer Science The University.
Ratbert: Nearest Sequence Memory Based Prediction Model Applied to Robot Navigation by Sergey Alexandrov iCML 2003.
Chinese Character Recognition for Video Presented by: Vincent Cheung Date: 25 October 1999.
General Algorithms for Common Business Problems
Revision (Part II) Ke Chen COMP24111 Machine Learning Revision slides are going to summarise all you have learnt from Part II, which should be helpful.
On Distinguishing between Internet Power Law B Bu and Towsley Infocom 2002 Presented by.
Image Subtraction for Real Time Moving Object Extraction Shahbe Mat Desa, Qussay A. Salih, CGIV’04.
Learning and Recognizing Activities in Streams of Video Dinesh Govindaraju.
Patterns and Sequences. Patterns refer to usual types of procedures or rules that can be followed. Patterns are useful to predict what came before or.
1 Activity and Motion Detection in Videos Longin Jan Latecki and Roland Miezianko, Temple University Dragoljub Pokrajac, Delaware State University Dover,
Face Alignment Using Cascaded Boosted Regression Active Shape Models
Math in Significant Figures. Significant Figures in Measurement and Calculations AAAA successful chemistry student habitually labels all numbers,
$100 $200 $300 $400 $500 $100 $200 $300 $400 $500 $100 $200 $300 $400 $500 $100 $200 $300 $400 $500 $100 $200 $300 $400 $500 $100 $200 $300.
A Study of Speed, Velocity and Acceleration
黃福銘 (Angus F.M. Huang) ANTS Lab, IIS, Academia Sinica TrajPattern: Mining Sequential Patterns from Imprecise Trajectories.
Neil Gealy 8/6/10. Pruning Plots What I have found to be the most optimal sequence of pruning: 1. Consecutive pruning based on radius and threshold 2.
Adding a Sequence of numbers (Pairing Method)
Motion – Part 1 Frame of Reference Distance vs. Displacement Average Speed.
8.1 and 8.2 answers. 8.3: Vectors February 9, 2009.
Fast calculation methods. Addition  Add 137,95 Solution: = (137-5)+100= = 232.
Accurate Video Localization Based on Google Maps Street View Accurate Video Localization Based on Google Maps Street View.
Presented by: Idan Aharoni
Ch. 11 Sec. 1 Distance & Displacement. Frame of Reference Describing motion accurately requires a Frame of Reference Describing motion accurately requires.
S-AIS Performance Analysis J.K.E. Tunaley London Research and Development Corporation, 114 Margaret Anne Drive, Ottawa, Ontario K0A 1L
Distance-Based Approaches to Inferring Phylogenetic Trees BMI/CS 576 Colin Dewey Fall 2010.
Cell Segmentation in Microscopy Imagery Using a Bag of Local Bayesian Classifiers Zhaozheng Yin RI/CMU, Fall 2009.
Overview Last two weeks we looked at evolutionary algorithms.
Algebra 1 Section 6.1 Solve and graph linear inequalities Inequalities Verbal Algebraically Graphically Inequality signs > greater than < less than > greater.
Speed Speed describes how fast an object is moving Speed describes how fast an object is moving If an object is not moving it has no speed If an object.
A guide for gathering data
Data Compression.
Motion.
Distance and Midpoint Formulas
Automation System For Checking Protein Prediction
Week 8 REU Nolan Warner.
Data Compression.
ASSIGNMENT NO.-2.
Clustering Uncertain Taxi data
Describing Motion.
Multiple Alignment and Phylogenetic Trees
The Functional Space of an Activity Ashok Veeraraghavan , Rama Chellappa, Amit Roy-Chowdhury Avinash Ravichandran.
Revision (Part II) Ke Chen
Physics Unit Introduction to Motion
Chapter 2: Reference Frames, Diplacement, and Velocity
Online Graph-Based Tracking
Revision (Part II) Ke Chen
Project P06441: See Through Fog Imaging
Arithmetic Sequences.
Unit 3 Review (Calculator)
Time Relaxed Spatiotemporal Trajectory Joins
Most of physics, and all of science, is concerned with motions.
Toward a Science of Computational Ethology
Finding Periodic Discrete Events in Noisy Streams
A guide for gathering data
Topological Signatures For Fast Mobility Analysis
INDOOR SCORING.
Calculate 9 x 81 = x 3 3 x 3 x 3 x 3 3 x 3 x 3 x 3 x 3 x 3 x =
The parallelogram and the isosceles triangle have the same perimeter.
Mathematics Unit 49: Moving Walkways
Progress Report Alvaro Velasquez.
Presentation transcript:

Accurate Video Localization Neil Gealy 7/20/10

Sequential Pruning

Sequential pruning Ground Truth in Green Numbers 1-5 represent frames that are kept after consecutive pruning 5 3 1 2 4

Sequential pruning Concept Frames which create a large change in total trajectory are probably out of sequence. Example: Frame 4 adds a large distance to the total trajectory compared to the other frames 5 3 1 2 4

Sequential pruning We determine the total length of the proposed trajectory after consecutive clustering Total Length = 100M for example 5 3 1 2 4

Sequential pruning Next, we remove one frame and calculate the total trajectory length. For example, removing frame 2. Total length = 85M 5 3 1 4

Sequential pruning For example, removing frame 3 Total Length = 80M 5 1 2 4

Sequential pruning For example, removing frame 4 Total Length = 70M 5 3 1 2

Sequential pruning We compare the new distances with the total length distance initially calculated. All frames = 100M Removing frame 1 = 78M Removing frame 2 = 85M Removing frame 3 = 80M Removing frame 4 = 70M Removing frame 5 = 72M By looking at the results, frame 4 has the most effect on the total trajectory length and is probably out of sequence so it is removed.

Sequential pruning Results after sequential pruning. (we removed frame 4 and renumbered) 4 3 1 2

Results Walking(1)

Sequential pruning – real example Results after consecutive and sequential pruning. Results after consecutive pruning.

Sequential pruning – real example Sequencing Average the numbers at each location. The average is assigned as the new sequence number for that location.

Sequential pruning – real example Sequencing To get the actual sequence, we look at each location (there are usually multiple matches for the same GPS location) We average the numbers at each location which represent the sequential ordering of the points. The average is assigned as the new sequence number for that location.

Results Walking(5)

Comparison Results after consecutive pruning Results after consecutive and sequential pruning

Sequential pruning – real example Sequencing Average the numbers at each location. The average is assigned as the new sequence number for that location.

Results Driving(1)

Sequential pruning – real example Results after consecutive and sequential pruning. Results after consecutive pruning.

Sequential pruning – real example Sequencing Average the numbers at each location. The average is assigned as the new sequence number for that location.

Results Walking(6)

Sequential pruning – real example Results after consecutive and sequential pruning. Results after consecutive pruning.

Sequential pruning – real example Sequencing Average the numbers at each location. The average is assigned as the new sequence number for that location.