Snimanje ljudskog tijela u pokretu i usporedba sustava OptiTrack i OpenPose

master's thesis

• Title (english): Human Body Motion Capture and Comparison of OptiTrack and OpenPose
• Author: Una Ikić
• Mentor: Tomislav Petković (mentor)
• Committee member: Tomislav Petković (predsjednik povjerenstva)
• Committee member: Davor Petrinović (član povjerenstva)
• Committee member: Tomislav Pribanić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-07-11, Croatia

Abstract

Snimanje ljudskog tijela u pokretu svoju primjenu pronalazi u raznolikim područjima poput sporta, medicine, rehabilitacije, nadzora, robotike, zabave i edukacije. U ovome radu uspoređuju se markerbased i markerless sustav za snimanje ljudskog pokreta. OptiTrack je točan i precizan sustav koji koristi markere, ali zahtijeva posebnu opremu i uvjete snimanja. OpenPose je sustav iz markerless skupine koji se temelji na dubokim neuronskim mrežama i koji prepoznaje ključne točke na tijelu. Osam kamera istodobno snimalo je četiri odabrane ljudske kretnje: sporo i brzo hodanje, skok prema naprijed i okret u mjestu. Nakon obrade snimljenih podataka pomoću oba sustava napravljena je usporedba x, y i z koordinata u vremenu za lijevo rame. Dobiveni rezultati pokazuju da trenutna implementacija OpenPose-a nije dovoljno dobra da zamijeni profesionalne sustave za praćenje ljudskog pokreta u mjerenjima koja zahtijevaju visoku preciznost. Razlike između dvaju sustava iznose do 37 centimetara, a postotak udaljenosti koje su manje od jednog centimetra za sve koordinate u prosjeku iznosi 21,73%, za udaljenosti manje od dva centimetra taj postotak iznosi 40,28%.

Abstract (english)

Human body motion capture finds applications in various fields such as sports, medicine, rehabilitation, surveillance, robotics, entertainment, and education. This paper compares marker-based and markerless systems for recording human movement. OptiTrack is an accurate and precise system that utilizes markers but requires special equipment and specific capturing conditions. OpenPose, on the other hand, is a markerless system based on deep neural networks that recognizes keypoints on the body. For the study, eight cameras simultaneously recorded four human movements: slow and fast walking, forward jump and turn in place. After processing the recorded data using both systems, a comparison of the x, y and z coordinates over time was conducted. The obtained results indicate that the current implementation of OpenPose is not sufficient to replace professional human motion tracking systems in measurements that require high precision. The differences between the two systems were found to be up to 97 centimeters. On average, the percentage of distances less than one centimeter for all coordinates was 21,73%, while the percentage of distances less than two centimeters was 40,28%.

• Keywords: OptiTrack; OpenPose; 3D; snimanje pokreta
• Keywords (english): OptiTrack; OpenPose; 3D; motion capture
• Language: croatian
• URN:NBN: urn:nbn:hr:168:232380
• Study programme: Title: Electrical Engineering and Information Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije (magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-10-11 09:04:48