Object Detection And Tracking Using Instant Online Feature Extraction
Journal Title: International Journal for Research in Applied Science and Engineering Technology (IJRASET) - Year 2015, Vol 3, Issue 5
Abstract
It is not always possible to provide labeled data for training because it requires substantial human effort, expensive tests, disagreement among experts. Labeling is not possible at instance level.To overcome these problem multiple instance learning (MIL) method is introduced which actively trained the data in online manner and combine with discriminative classifier which separate the object from its background and provide positive and negative bags. The fisher information criteria is used to train dataset in online manner which perfectly describe the label of positive content in positive label bag and negative content in negative label bags. The use of actively trained classifier helps to improve the efficiency of tracking object in motion.
Authors and Affiliations
Mr. Suraj R. Jaronde
Keywords
Related Articles
- EP ID EP20341
- DOI -
- Views 260
- Downloads 4
slugPower Efficient Location Based Routing
In wireless ad - hoc networks, there are several characteristics different among wired networks. The differences are varying of network topology, limited resources like bandw idth and energy...
Performance Analysis of Adaptive Modulation in OFDM Systems
OFDM technology promises to be a key technique for achieving the high data capacity and spectral efficiency requirements for wireless communication systems. Different order modulations allows a wireless communication sy...
Smart Wireless Attendance Monitoring Using NFC
With the ever growing trends in technology its has become essential to develop smart systems in each field which give more output in less involvement of man power. This paper represents one such idea of Smart Attendance...
Data Compression Technique for Cloud Storage in Multi-User Environments
In many associations, the cloud storages contain a copy (duplicates) of information. For example, the same file would be spared in various locations by different users, too many documents that aren’t differentiable may...
Implementation of Embedded System Based Face Recognition System
Face Recognition is the ability to detect and recognize a person by their facial characteristics. Face is a multidimensional and hence requires a lot of mathematical computations. Face recognition system is very essenti...