Scribd
Upload
125 views13 pages

Pedestrian Detection and Tracking

This document discusses pedestrian detection and tracking. It begins by outlining the need for pedestrian detection in applications like advanced driver assistance systems, CCTV surveillance, and assisting visually impaired people. It then defines the problem statement of detecting and tracking pedestrians as part of an advanced driver assistance system. The document surveys various solutions in the literature, including part-based detection, patch-based detection, motion-based detection, and detection using HOG descriptors with SVM. It compares these solutions and identifies using a particle filter in MATLAB as the best solution for pedestrian tracking.

Uploaded by

dfghjkl
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
125 views13 pages

Pedestrian Detection and Tracking

This document discusses pedestrian detection and tracking. It begins by outlining the need for pedestrian detection in applications like advanced driver assistance systems, CCTV surveillance, and assisting visually impaired people. It then defines the problem statement of detecting and tracking pedestrians as part of an advanced driver assistance system. The document surveys various solutions in the literature, including part-based detection, patch-based detection, motion-based detection, and detection using HOG descriptors with SVM. It compares these solutions and identifies using a particle filter in MATLAB as the best solution for pedestrian tracking.

Uploaded by

dfghjkl
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 13

PEDESTRIAN DETECTION AND

TRACKING

Guided by: Presented By:


Dr. Nalini C Iyer Shreya Bhat : 2BV14EC098
Spoorthi S.K : 2BV14EC107
Vinuta Kadiwal : 2BV14EC123
Contents
Need gathering
Identifying problem
Understanding of professional ethics Copy right, plagiarism
Problem definition
Literature survey
Multiple Solution
Best Solution recognised
NEED GATHERING
Switch over from manual to automatic locomotives.
In CCTV cameras for tracking suspicious people/criminals
Location and motion trajectories is necessary for:
1.Safe Navigation
2.Path Planning
3.Collision Prevention
An aid to visually impaired people.
Pedestrian Counting
Traffic and Pedestrian Control.
Traffic Flow Pattern identification.
Miscellaneous:
1.Tracking game of footballers.
2.Virtual Reality
3. Performance measurement of athletes require that certain body parts
tracked.
IDENTIFYING PROBLEM
PROBLEM STATEMENT

There is need for pedestrian detection and tracking as a part


of Advanced Driver Assistance System , to aid driverless cars.
PATENT SEARCH
Indoor/outdoor pedestrian navigation
US 8972182 B1
Inventors: Eric Foxlin

ABSTRACT
Among other things, positioning a magnetic instrument on a pedestrian;
positioning an inertial instrument on a foot of a pedestrian; receiving
positioning signals at the pedestrian; aligning the inertial instrument based
in part on the received positioning signals; calibrating the magnetic
instrument using the inertial instrument; and tracking the pedestrian using
the calibrated magnetic instrument and the inertial instrument.

https://www.google.co.in/patents/US8972182
Image and video based pedestrian traffic estimation
US 20130101159 A1
Inventors: Hui Chao, Rajarshi Gupta
ABSTRACT
Person detection and tracking techniques may be used to estimate
pedestrian traffic in locations equipped with cameras.
Persons detected in video data from the cameras may help determine
existing pedestrian traffic data.
Future pedestrian traffic estimation may be performed to estimate
pedestrian traffic characteristics (such as volume, direction, etc.) Such
traffic estimation may be provided to users for route
planning/congestion information.
A traffic map can be derived based on the number of people at or
expected to be at certain locations. The map may be provided to users
to provide traffic data and/or estimations
Pedestrian Tracking by Associating Tracklets using Detection Residuals

ABSTRACT:
Due to increased interest in visual surveillance, various multiple object
tracking methods have been recently proposed and applied to pedestrian
tracking.
However in presence of intensive inter-object occlusion and sensor gaps,
most of these methods result in tracking failures.
A two-stage multi-object tracking approach to robustly track pedestrians in
such complex scenarios.
It first generates high confidence partial track segments (tracklets) using a
robust pedestrian detector and then associate the tracklets in a global
optimization framework.
Unlike the existing two-stage tracking methods, the method uses the un-
associated low confidence detections (residuals) between the tracklets,
which improves the tracking performance.
The evaluation of the method on the CAVIAR dataset and show that the
method performs better than state-of-the-art methods.
LITERATURE SURVEY
http://www.cs.stanford.edu/~amirz/index_files/PED12_v2.pdf : This paper is about
detection of low density and high density crowd using PATCH based detection using
HOG descriptors.
http://www-users.cs.umn.edu/~masoud/publications/masoud-pedestrian-itsc-1997.pdf
http://iris.usc.edu/outlines/papers/2008/singh-wu-nevatia-motion08.pdf : The paper
explains about the tracklet based pedestrian detection.
Limitation: One limitation of the current system is that we use tracklets for track
initiation and hence any track for which no tracklet is generated cannot be tracked by
this method.
Another limitation is that the method does not fix the identity switches that occur
within the tracklets. We intend to address these issues in future.
Algorithms based of Kalman and Particle filter
Different Software used for pedestrian detection are:
Open CV
Python
Matlab
HOG
LITERATURE SURVEY
HoG ( Histogram of Oriented Graphics) and SVM
In this local object appearance and shape of an image can be described
by the distribution of intensity gradients or edge directions.
The image is divided into cells, and for the pixels within each cell, a
histogram of gradient directions is compiled. The descriptor is the
concatenation of these histograms.
For improved accuracy, the local histograms can be contrast-normalized
by calculating a measure of the intensity across a larger region of the
image, called a block.
The HOG descriptor has a few key advantages over other descriptors.
As cell concept- Invariant to geometric and photometric
transformations, except for object orientation.
Coarse spatial sampling
Fine orientation sampling
Strong local photometric normalization.
MULTIPLE SOLUTION IDENTIFIED FOR
DETECTION
Holistic Detection
Edge Detection
HOG Descriptors and SVM
Part-based Detection
Patch-based Detection
Implicit Shape Model
Motion-based Detection
Detection using multiple cameras.
MULTIPLE SOLUTIONS IDENTIFIED FOR
TRACKING
Kalman Filter
Particle Filter
COMPARISON OF SOLUTIONS
S.NO Solution Identified Limitation/Advantages

1 Part-based Detection Feature Selection

2 Patch-based Detection 1.Similar results for varying environments.


2.Desciptors are not enough distinctive under
image noise

3 Motion-based Detection 1.No accuracy due to fixed camera


2.Far objects/pedestrians cant be detected.

4 HoG and SVM ** 1.Cell Concept: invariant to geometric and


photometric variations.
2. Coarse Spatial Sampling
3.Fine Orientation Sampling
5 Detection using Multiple Camera Expensive, Complex algorithms and mechanisms
SOLUTION IDENTIFIED

Pedestrian Tracking using Particle Filter , using MATLAB

You might also like