Skeleton Tracking Based Complex Human Activity Recognition Using Kinect Camera

Anjum, Muhammad Latif; Ahmad, Omar; Rosa, Stefano; Yin, Jingchun; Bona, Basilio

doi:10.1007/978-3-319-11973-1_3

Cited by 9 publications

(9 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We mounted a Microsoft Kinect on Baxter, and used OpenNITracker to find human joint positions. This depth image based skeleton tracking approach produces robust joint positions for different users in various background and lighting conditions [1]. Baxter makes repositioning requests by displaying text on its screen.…”

Section: Discussionmentioning

confidence: 99%

Personalized Assistance for Dressing Users

Klee

Ferreira

Silva

et al. 2015

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. In this paper, we present an approach for a robot to provide personalized assistance for dressing a user. In particular, given a dressing task, our approach finds a solution involving manipulator motions and also user repositioning requests. Specifically, the solution allows the robot and user to take turns moving in the same space and is cognizant of the user's limitations. To accomplish this, a vision module monitors the human's motion, determines if he is following the repositioning requests, and infers mobility limitations when he cannot. The learned constraints are used during future dressing episodes to personalize the repositioning requests. Our contributions include a turn-taking approach to humanrobot coordination for the dressing problem and a vision module capable of learning user limitations. After presenting the technical details of our approach, we provide an evaluation with a Baxter manipulator.

show abstract

Section: Discussionmentioning

confidence: 99%

Personalized Assistance for Dressing Users

Klee

Ferreira

Silva

et al. 2015

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…A Kinect is a sensor as other sensor have its advantages and disadvantages [11,12,13,16,[18][19][20][21] . Many authors and programmer make the Kinect camera their first choice because it has been yet the first camera which combines all the following advantages:…”

Section: Why the Kinectmentioning

confidence: 99%

“…The above hardware features make the Kinect able to provide full-body 3D motion capture, facial recognition, and voice recognition capabilities [17]. Which make the Kinect very useful in various application such as: object tracking and recognition [19,21,22], human skeleton tracking and activity analysis [16,21,25], hand gesture analysis [16,26], 3D-simultaneous localization and mapping [19,22], emergences detection such as: assault detection [27,28] and fall detection [11,13,29], and other.…”

Section: Kinect Hardwarementioning

confidence: 99%

Computer Vision Based Fall Detection Methods Using the Kinect Camera : A Survey

Jarraya

2018

IJCSIT

View full text Add to dashboard Cite

Disabled people can overcome their disabilities in carrying out daily tasks in many facilities [1]. However, they frequently report that they experience difficulty being independently mobile. And even if they can, they are likely to have some serious accidents such as falls. Furthermore, falls constitute the second leading cause of accidental or injury deaths after injuries of road traffic which call for efficient and practical/comfortable means to monitor physically disabled people in order to detect falls and react urgently. Computer vision (CV) is one of the computer sciences fields, and it is actively contributing in building smart applications by providing for image\video content "understanding." One of the main tasks of CV is detection and recognition. Detection and recognition applications are various and used for different purposes. One of these purposes is to help of the physically disabled people who use a cane as a mobility aid by detecting the fall. This paper surveys the most popular approaches that have been used in fall detection, the challenges related to developing fall detectors, the techniques that have been used with the Kinect in fall detection, best points of interest (joints) to be tracked and the well-known Kinect-Based Fall Datasets. Finally, recommendations and future works will be summarized.

show abstract

“…Our dataset is purposely built to be used with ROS based OpenNI skeleton tracker and should serve as a reference for future research in Kinect based activity recognition. The dataset was initially presented in [1] and we have now added few more activities to the dataset. The dataset now consists of a total of 10 activities being performed by different users.…”

Section: Creating the Datasetmentioning

confidence: 99%

“…This not only reduces the computational complexity of the algorithm but also enhances the recognition accuracy as shown in experimental results. In the previous work ( [1,2]), we presented a method for skeleton tracking based activity classification based on the manual selection of subset joints depending upon the activity. With this paper, we extend that approach to automatic selection of subset joints most relevant to the activity.…”

Section: Introductionmentioning

confidence: 99%

Tracking a Subset of Skeleton Joints: An Effective Approach towards Complex Human Activity Recognition

Anjum

Rosa

Bona

2017

Journal of Robotics

Self Cite

View full text Add to dashboard Cite

We present a robust algorithm for complex human activity recognition for natural human-robot interaction. The algorithm is based on tracking the position of selected joints in human skeleton. For any given activity, only a few skeleton joints are involved in performing the activity, so a subset of joints contributing the most towards the activity is selected. Our approach of tracking a subset of skeleton joints (instead of tracking the whole skeleton) is computationally efficient and provides better recognition accuracy. We have developed both manual and automatic approaches for the selection of these joints. The position of the selected joints is tracked for the duration of the activity and is used to construct feature vectors for each activity. Once the feature vectors have been constructed, we use a Support Vector Machines (SVM) multiclass classifier for training and testing the algorithm. The algorithm has been tested on a purposely built dataset of depth videos recorded using Kinect camera. The dataset consists of 250 videos of 10 different activities being performed by different users. Experimental results show classification accuracy of 83% when tracking all skeleton joints, 95% when using manual selection of subset joints, and 89% when using automatic selection of subset joints.

show abstract

Skeleton Tracking Based Complex Human Activity Recognition Using Kinect Camera

Cited by 9 publications

References 12 publications

Personalized Assistance for Dressing Users

Personalized Assistance for Dressing Users

Computer Vision Based Fall Detection Methods Using the Kinect Camera : A Survey

Tracking a Subset of Skeleton Joints: An Effective Approach towards Complex Human Activity Recognition

Contact Info

Product

Resources

About