magistrsko delo
Abstract
Segmentacija telesa je proces označevanja delov slike in njihova klasifikacija v semantične razrede. Segmentacija telesa se uporablja v mnogih aplikacijah in sistemih, ki obdelujejo slike ljudi, ter predstavlja enega izmed ključnih delov takšnih cevovodov. V tem delu se ukvarjamo z izboljšavo obstoječih modelov za segmentacijo ljudi in pri tem uporabimo metodo večciljnega učenja. Predstavimo in implementiramo večciljni model SPD, ki poleg segmentacije vključuje tudi naloge napovedi telesnega skeleta in napoved globinske predstavitve telesa. Implementiramo tudi modele, ki poleg segmentacije vključujejo le eno izmed nalog. Pripravimo zbirko slik, ki vsebuje vse anotacije, ki so potrebne za učenje modela, in modele z njimi naučimo. Rezultate vseh implementiranih modelov analiziramo in primerjamo z referenčnimi modeli JPPNet in DensePose. Uporabljene metrike segmentacijske kakovosti kažejo na izboljšavo vseh metrik v primerjavi z modelom JPPNet. Rezultati skeletne in globinske predstavitve so glede na referenčne modele nekoliko slabši.
Keywords
segmentacija;razločanje delov telesa;večciljno učenje;magisteriji;
Data
Language: |
Slovenian |
Year of publishing: |
2021 |
Typology: |
2.09 - Master's Thesis |
Organization: |
UL FRI - Faculty of Computer and Information Science |
Publisher: |
[J. Jug] |
UDC: |
004.93(043.2) |
COBISS: |
91319043
|
Views: |
239 |
Downloads: |
51 |
Average score: |
0 (0 votes) |
Metadata: |
|
Other data
Secondary language: |
English |
Secondary title: |
Body segmentation using multi-task learning |
Secondary abstract: |
Body segmentation is a process of labelling parts of images and classifying them into semantic classes of body parts and clothing pieces. Body segmentation is used in many complex systems that process images of people and is one of the key parts of such processing pipelines. In this work, we try to improve the existing models for human segmentation using multi-task learning method. We present and implement a multi-task model SPD, which, in addition to segmentation, also includes the tasks of predicting the body skeleton and the prediction of dense pose. We also implement other models that include only one additional task next to the segmentation task. We prepare a data set of images that contain all three necessary annotations types for learning. The results of the implemented models are analysed and compared with the results of JPPNet reference model and DensePose model. The Segmentation results indicate an improvement in all metrics in comparison to the JPPNet segmentation model. The results of skeletal and dense pose representations perform a little worse than the reference models. |
Secondary keywords: |
computer vision;segmentation;human body parsing;multi task learning;computer science;master's degree;Računalniški vid;Računalništvo;Univerzitetna in visokošolska dela; |
Type (COBISS): |
Master's thesis/paper |
Study programme: |
1000471 |
Thesis comment: |
Univ. v Ljubljani, Fak. za računalništvo in informatiko |
Pages: |
65 str. |
ID: |
14092807 |