Geri Dön

A study on human activity recognition with analysis of angles between skeletal joints

Başlık çevirisi mevcut değil.

  1. Tez No: 622884
  2. Yazar: ÖMER FARUK İNCE
  3. Danışmanlar: PROF. DR. DANIŞMAN YOK
  4. Tez Türü: Doktora
  5. Konular: Elektrik ve Elektronik Mühendisliği, Electrical and Electronics Engineering
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 2018
  8. Dil: İngilizce
  9. Üniversite: Kyungsung University
  10. Enstitü: Yurtdışı Enstitü
  11. Ana Bilim Dalı: Belirtilmemiş.
  12. Bilim Dalı: Belirtilmemiş.
  13. Sayfa Sayısı: 90

Özet

Human activity recognition (HAR) has become effective in computer vision for video surveillance systems and medical purposes such as monitoring elder people in home environment. There are various types of sensors which can be used to address this task. However, RGB-depth sensors, mostly the ones used for gaming purposes, are cost effective and enable to provide data about environment. In this thesis, a biometric system that can detect human activities in 3D space is proposed. Previous approaches have shown that monitoring of various physiological signals can provide distinctive information to recognize human activities. The proposed method conducts a machine learning to determine a pattern on numerous activities using angles between joints in 3D space. Simply, activity related skeletal joint angles are obtained using RGB-depth sensor. Since HAR is operated in a time domain, swinging angle information is stored by sliding kernel method which creates a feature set. A Haar-wavelet transform (HWT) then is applied not to lose the important information in the signal before reducing dimension. After that, dimension reduction using averaging algorithm is also applied to decrease number of features and computational cost. Before the classification, proposed inverse HWT is applied. By this means, new feature set is extracted. Lastly, k-nearest neighbor (k-nn) algorithm is used to recognize the activity with respect to the given data. Since k-nn is lazy learning algorithm, computational cost is quite low compare to other machine learning algorithms. The performance of the proposed method for various activities show that human activity recognition rate is 86.1%. To see the effectiveness of the proposed method, there are two types of comparisons; firstly, k-nn performances based on different 'k' value are shown. Secondly, test performance is observed based on different classifiers. It is expected that proposed algorithm can make a contribution for improvement of the HAR

Özet (Çeviri)

Özet çevirisi mevcut değil.

Benzer Tezler

  1. Le rapport des droits de l'homme au politique: Lefort et Rancière

    İnsan haklarının politik-olan bağlantısı: Lefort ve Rancière

    EYLEM YOLSAL MURTEZA

    Doktora

    Fransızca

    Fransızca

    2022

    FelsefeGalatasaray Üniversitesi

    Felsefe Ana Bilim Dalı

    PROF. DR. ALİYE KARABÜK KOVANLIKAYA

  2. Bebeklerin beden dilini poz tabanlı gerçek zamanlı çözme

    Pose-based real-time decoding of babies' body language

    MUHAMMET MÜCAHİT ENES YURTSEVER

    Yüksek Lisans

    Türkçe

    Türkçe

    2022

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve KontrolKocaeli Üniversitesi

    Bilişim Sistemleri Mühendisliği Ana Bilim Dalı

    DOÇ. DR. SÜLEYMAN EKEN

  3. Cihaz sensörleri ile insan aktivitelerine dayalı makine öğrenmesi

    Machine learning based on human activities with device sensors

    MUHAMMED YUSUF UYSAL

    Yüksek Lisans

    Türkçe

    Türkçe

    2024

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Aydın Üniversitesi

    Bilgisayar Mühendisliği Ana Bilim Dalı

    DR. ÖĞR. ÜYESİ SİNA APAK

  4. A comparative study on human activity classification with miniature inertial and magnetic sensors

    Minyatür eylemsizlik duyucuları ve manyetometreler ile insan aktivitelerinin sınıflandırılması üzerine karşılaştırmalı araştırma

    MURAT CİHAN YÜKSEK

    Yüksek Lisans

    İngilizce

    İngilizce

    2011

    Elektrik ve Elektronik Mühendisliğiİhsan Doğramacı Bilkent Üniversitesi

    Elektrik ve Elektronik Mühendisliği Bölümü

    PROF. DR. BİLLUR BARSHAN

  5. Human activity classification with miniature inertial sensors

    Minyatür eylemsizlik duyucuları kullanılarak insan hareketlerinin sınıflandırılması

    ORKUN TUNÇEL

    Yüksek Lisans

    İngilizce

    İngilizce

    2009

    Elektrik ve Elektronik Mühendisliğiİhsan Doğramacı Bilkent Üniversitesi

    Elektrik ve Elektronik Mühendisliği Bölümü

    PROF. DR. BİLLUR BARSHAN