izpis_h1_title_alt

Avtomatsko merjenje dolžin smučarskih skokov z globokimi nevronskimi mrežami
ID Nabergoj, David (Author), ID Kukar, Matjaž (Mentor) More about this mentor... This link opens in a new window

.pdfPDF - Presentation file, Download (11,23 MB)
MD5: C4505D2D0A7AC0FCAD1E871AF88D94A6

Abstract
V diplomski nalogi obravnavamo problem avtomatskega merjenja dolžin smučarskih skokov na podlagi videoposnetkov. Postopek razdelimo na dva podproblema: napovedovanje trenutka doskoka in določanje dolžine skoka. Prvega rešujemo s konvolucijsko nevronsko mrežo, ki za dano sličico videoposnetka skoka napove, ali je skakalec v zraku ali na tleh. Dolžino skoka določimo z uporabo klasičnih metod računalniškega vida, s katerimi najprej poiščemo točko stopal na sliki, nato pa z upoštevanjem oddaljenosti točke od merilnih črt pridobimo natančno dolžino. Konvolucijska nevronska mreža doseže klasifikacijsko točnost 93 %, celoten postopek določanja dolžine skoka pa srednjo absolutno napako 0.785 metra na relevantnem območju doskočišča. Napovedan trenutek doskoka se od resničnega razlikuje za približno eno sličico. Rezultati diplomske naloge pomenijo prispevek k razvoju sodobnih sistemov za avtomatsko meritev dolžin smučarskih skokov v realnem času.

Language:Slovenian
Keywords:smučarski skoki, video meritve, meritve dolžin, globoko učenje, računalniški vid
Work type:Bachelor thesis/paper
Organization:FRI - Faculty of Computer and Information Science
Year:2019
PID:20.500.12556/RUL-109332 This link opens in a new window
COBISS.SI-ID:1538309315 This link opens in a new window
Publication date in RUL:30.08.2019
Views:1870
Downloads:474
Metadata:XML RDF-CHPDL DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:English
Title:Automatic ski-jump distance measurement with deep neural networks
Abstract:
We consider the problem of automatic video-based ski-jump distance measurement. The procedure is split into two subproblems: predicting the landing and determining the distance of the jump. To predict the landing, we use a convolutional neural network which takes an image of the ski-jump video as input and predicts whether the ski-jumper is in the air or on the ground. To determine the distance of the jump, we use classical computer vision methods which first find the location of the jumper's feet in the image and then use measurement lines to output the precise distance. The convolutional neural network achieves a classification accuracy of 93%. The complete procedure achieves a mean absolute error of 0.785 meters in the relevant landing area. The predicted landing and the actual landing differ by approximately one frame. The results of the thesis contribute to the development of modern real-time ski-jump distance measurement systems.

Keywords:ski-jumping, video measurement, distance measurement, deep learning, computer vision

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back