Mоlimо vаs kоristitе оvај idеntifikаtоr zа citirаnjе ili оvај link dо оvе stаvkе:
https://open.uns.ac.rs/handle/123456789/1712
Nаziv: | Deep learning powered automated tool for generating image based datasets | Аutоri: | Arsenović, Marko Sladojević, Srđan Anderla, Andraš Stefanović, Darko Lalić, Bojan |
Dаtum izdаvаnjа: | 27-мар-2018 | Čаsоpis: | 2017 IEEE 14th International Scientific Conference on Informatics, INFORMATICS 2017 - Proceedings | Sažetak: | © 2017 IEEE. During the last decade, rapid growth of the social media usage and the Internet in general led to the drastic expansion of the publicly available data. This gives the great opportunity for research and analysis in many fields of machine learning and data science. Even though this data is open to use, often it is distributed over the various sources on the web and in order to be used in data science experiments it requires to be gathered, properly labeled and pre-processed. This is the essential step of every research and it is generally time consuming. In this paper, the automated system for gathering and preparing the dataset is proposed. The system consists of numerous features which embrace cutting edge approaches in order to reduce the time for developing the dataset. It is primarily developed for image based datasets, and provides possibility for researchers to collect the images from the different sources: social media by official APIs or online image libraries using web scraping techniques. Along with the data gathering, proposed system provides an intelligent image labeling feature by using state-of-the-art deep learning methods which use trained models for object recognition and label them in standardized formats. The system is modular and could be extended for desired objects detection by simply adding new trained models. Finally, it provides the possibility to speed up the researchers by setting their focus more on the essential parts of their studies than to generate the dataset by themselves. | URI: | https://open.uns.ac.rs/handle/123456789/1712 | ISBN: | 9781538608890 | DOI: | 10.1109/INFORMATICS.2017.8327214 |
Nаlаzi sе u kоlеkciјаmа: | FTN Publikacije/Publications |
Prikаzаti cеlоkupаn zаpis stаvki
SCOPUSTM
Nаvоđеnjа
3
prоvеrеnо 03.05.2024.
Prеglеd/i stаnicа
51
Prоtеklа nеdеljа
22
22
Prоtеkli mеsеc
2
2
prоvеrеnо 10.05.2024.
Google ScholarTM
Prоvеritе
Аlt mеtrikа
Stаvkе nа DSpace-u su zаštićеnе аutоrskim prаvimа, sа svim prаvimа zаdržаnim, оsim аkо nije drugačije naznačeno.