In this project, we deal with a central challenge that is to make egocentric
videos watchable. First person videos are generally long-running streams with
unedited content, which make them boring and visually unpalatable. In this
project we propose a novel methodology to compose the new fast-forward video by
selecting frames in a smart manner and based in the semantic information
extracted from images.
Compared with daily human gestures, moves in high performance sports are
faster and have low inter-class variability, which produce noisy features and
ambiguity. We present a discriminative key pose-based approach for moves
recognition and segmentation of training sequences for high performance sports.
Efficient 3D model estimation methodology for large datasets
The advent of digital cameras heralded many possibilities of structure and shape re-
covery from imagery that are quickly and inexpensively acquired by such devices.
In this work we propose an efficient approach based on structure from motion and
multi-view stereo reconstruction techniques to automatically generate DEM - Digital
Elevation Models - from aerial images and also 3D models in general.
Image and Depth Super-Resolution
With the purpose of increasing data
resolution, at the same time reducing noise and filling the holes
in the depth maps, in this work we propose a method that
combines depth fusion and image reconstruction in a super-
Download the conference paper (Best Computer Graphics Paper Award of SIBGRAPI 2015).
Keypoint descriptor extraction for RGB-D data
At the core of a myriad of tasks such as object recognition, tridimensional recon-
struction and alignment resides the critical problem of correspondence. We
introduce BRAND descriptors that efficiently combine appearance and geometrical
shape information from RGB-D images, and are largely invariant to rotation,
illumination changes and scale transformations.