| Grant number: | 13/27101-1 |
| Support Opportunities: | Scholarships in Brazil - Doctorate |
| Start date: | September 01, 2014 |
| End date: | April 30, 2018 |
| Field of knowledge: | Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques |
| Agreement: | Coordination of Improvement of Higher Education Personnel (CAPES) |
| Principal Investigator: | Agma Juci Machado Traina |
| Grantee: | Letrícia Pereira Soares Avalhais |
| Host Institution: | Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil |
| Associated scholarship(s): | 16/10703-7 - Video event recognition with deep learning, BE.EP.DR |
Abstract In a scenario of intense generation and dissemination of multimedia data, a growing demand for technology processing, compression and retrieval for these data, especially digital videos is observed. Accuracy and efficiency are critical factors to enable methods of storing and managing applications of large volumes of video data. Several Content- based Video Retrieval Systems - (CBVR) have been proposed, however, the support to access videos on a semantic level remains a challenge due to the problem of semantic gap. Such systems generally use video segmentation as a unit for indexing and retrieval, so the quality of segmentation is directly reflected in the performance of these systems. The use of multi-modal features, visual, motion, and audio, for example, when analyzed together with the methods of finding patterns in time series can provide significant gain on representativeness of these features that are highly dependent on the semantic content of the data. This project aims to develop algorithms and methods for summarization and video segmentation, to remove noisy and irrelevant segments and get a representation more representative for the semantic context content of the video through temporal exploration of association between characteristics of different modalities, improving its content retrieval. The contributions of this project will be evaluated in the domain of films and television series in the context of content-based video retrieval systems, as well as data obtained via Crowdsourcing from the RESCUER project for incident detection in emergency situations. (AU) | |
| News published in Agência FAPESP Newsletter about the scholarship: | |
| More itemsLess items | |
| TITULO | |
| Articles published in other media outlets ( ): | |
| More itemsLess items | |
| VEICULO: TITULO (DATA) | |
| VEICULO: TITULO (DATA) | |