TeachingSs08ActionRecognition

Chair for Computer Aided Medical Procedures & Augmented Reality
Lehrstuhl für Informatikanwendungen in der Medizin & Augmented Reality

Hauptseminar: Action Recognition and Analysis: From Signals to Models

Nassir Navab, Nicolas Padoy, Tobias Blum

SWS: 2+0
ECTS: 4 Credits
Time/Location: On thursdays, MI 03.13.010 - Time to be announced, probably block-wise
The seminar language is English.

Registration

If you are interested in taking part in the seminar

  • please send an email to Tobias Blum()
  • initial meeting: date will be announced

Please include last and first name, matriculation number, current semester, discipline (diploma, master, bachelor) including major, and your favorite topic.

Introduction

Especially in highly complex environments, like the operating room, the amount of sensors is steadily increasing. This includes biomedical data like heart beat and blood pressure of the patient, human motion data obtained by using accelerometers, optical and magnetical tracking systems. Other signals include position and orientation of tools, also obtained using tracking systems, and signals derived from video images using computer vision techniques. An interesting field of research for computer scientists is to recognize and analyze human actions based on these signals. Possible applications include context-sensitive user interfaces, monitoring, human understandable representation of workflows, training and evaluation. This seminar will cover all related topics from low level signal analysis up to building high level models of human actions. The seminar will deal mainly, but not exclusively, with medical applications.

WorkflowLaparoscopy SurgeonTracking WorkflowVertebroplasty

Topics

  • Sensor and signals
    • Optical and magnetical tracking systems
    • Interial sensors like accelerometers and gyroscopes
    • RFID
  • Signal processing
    • Computer Vision to derive signals from video images
    • Motif discovery techniques to find recurring pattern
    • Gesture recognition
  • Modeling of actions and workflows
    • Hidden Markov Models
    • Dynamic Time Warping
    • Dynamic Bayes Nets
  • Applications
    • Context-sensitive user interfaces
    • Use for training and evaluation
    • Monitoring

You are interested in a topic that does not appear here? Feel free to send us an email.

Literature

K. Cleary, H.Y. Chung, S. K. Mun. OR 2020: The operating room of the future, Laparoscopic and Advanced Surgical Techniques, 2005
Y. Shi, Y. Huang, D. Minnen, A. Bobick, I. Essa. Propagation Networks for Recognition of Partially Ordered Sequential Actions, CVPR, 2004
D. Minnen, I. A. Essa, T. Starner: Expectation Grammars: Leveraging High-Level Expectations for Activity Recognition, CVPR, 2003
A. Ahmadi, T. Sielhorst, R. Stauder, M. Horn, H. Feußner, N. Navab. Recovery of surgical workflow without explicit models, MICCAI,2006
N. Padoy, T. Blum, I. Essa, H. Feußner, M.O. Berger, N. Navab. A Boosted Segmentation Method for Surgical Workflow Analysis, MICCAI 2007
T. Mackel, J. Rosen, C. Pugh. Application of Hidden Markov Modeling to Objective Medical Skill Evaluation, MMVR 2007
R. Leuschke, A. Bhandari, B. Sires, B. Hannaford. Low Cost Eye Surgery Simulator with Skill Assessment Component, MMVR 2007
A. James, D. Vieira, Z. Yang , A. Darzi. Eye-Gaze Driven Surgical Workflow Segmentation, MICCAI 2007
K. Kahol, N. Krishnan, V. Balasubramanian, S. Panchanthan, M. Smith, J. Ferrara. Measuring movement expertise in surgical tasks, MM 2006
R. Hamid, S. Maddi, A. Bobick, I. Essa. Structure from Statistics - Unsupervised Activity Analysis using Suffix Trees, ICCV 2007
D. Minnen, C. Isbell, I. Essa, T. Starner. Discovering Multivariate Motifs using Subsequence Density Estimation, AAAI 2007
B. Bhatia, T. Oates, Y. Xiao, P. Ho. Real-Time Identification of Operating Room State from Video, CAS 2007
D. Kwon, M. Gross. A Framework for 3D Spatial Gesture Design and Modeling Using a Wearable Input Device, ISWE 2007
D. Weinland, E. Boyer, R. Ronfard. Action Recognition from Arbitrary Views using 3D Exemplars, ICCV 2007


Useful Links

Requirements

  • the presentation should be about 45min (here is a Power Point template)
  • the handout of about 10 pages (in LaTex, here is a template)
  • the outline should be sent to the supervisor one month before the presentation
  • the presentation slides and handout should be finalized and discussed with the supervisor one week before the presentation here is a template

TeachingForm
Title: Action Recognition and Analysis: From Signals to Models
Professor: Nassir Navab
Tutors: Nicolas Padoy, Tobias Blum
Type: Hauptseminar
Information: 2 SWS, 4 ECTS
Term: 2008SoSe
Abstract: This seminar covers recognition and analysis of human actions based on signals obtained from different sesnors.


Edit | Attach | Refresh | Diffs | More | Revision r1.5 - 05 Jan 2009 - 13:03 - NicolasPadoy

Lehrstuhl für Computer Aided Medical Procedures & Augmented Reality    rss.gif