Hierarchical Least Square Twin Support VectorMachines Based Framework for Human ActionRecognition
Publish Year: 1390
نوع سند: مقاله کنفرانسی
زبان: English
View: 1,411
This Paper With 5 Page And PDF Format Ready To Download
- Certificate
- من نویسنده این مقاله هستم
استخراج به نرم افزارهای پژوهشی:
شناسه ملی سند علمی:
ICMVIP07_109
تاریخ نمایه سازی: 28 مرداد 1391
Abstract:
The aim of this paper is presentation of a new humanaction recognition framework. In the proposed framework, localspace-time features extracted by use of Harris detector algorithmand Histogram of Optical Flow (HOF). A new classifier based ontwo non-parallel hyperplanes called Twin Support VectorMachines (TWSVM) is used which is four times faster thanclassical SVM. According to the prior knowledge that two classesof human action recognition (jogging and running) are verysimilar and recognition of these classes are difficult, ahierarchical structure is used for better recognition. We appliedour method to KTH dataset to investigate the performance of theproposed action recognition approach. Our experimental resultshown that our approach improves state-of-the-art results byachieving 98.33%, 96.39% in case of leave-one-out and 10-foldcross validation.
Keywords:
Action Recognition , Twin Support Vector Machines , Histogram of Optical Flow (HOF) , Harris , PCA , KTH dataset ,
Authors
Kourosh Mozafari
Department of Electrical and computer EngineeringTarbiat Modares UniversityTehran, Iran
Nasrollah Moghadam Charkari
Department of Electrical and computer EngineeringTarbiat Modares UniversityTehran, Iran
Jalal A. Nasiri
Department of Electrical and computer EngineeringTarbiat Modares UniversityTehran, Iran
Saeed Jalili
Department of Electrical and computer EngineeringTarbiat Modares UniversityTehran, Iran
مراجع و منابع این Paper:
لیست زیر مراجع و منابع استفاده شده در این Paper را نمایش می دهد. این مراجع به صورت کاملا ماشینی و بر اساس هوش مصنوعی استخراج شده اند و لذا ممکن است دارای اشکالاتی باشند که به مرور زمان دقت استخراج این محتوا افزایش می یابد. مراجعی که مقالات مربوط به آنها در سیویلیکا نمایه شده و پیدا شده اند، به خود Paper لینک شده اند :