m. borges, 2 tiago a. fonseca, 3 ricardo l. de queiroz 4 and touradj ebrahimi 1Recent trends in multimedia technologies indicate the need for richer imaging modalities to increase user engagement with content.Among other alternatives, point clouds denote a viable solution that offers an immersive content representation, as witnessed by current activities JPEG MPEG standardization committees.As result of such efforts, is at final stages drafting emerging standard cloud compression, which we...

We present a method to super-resolve voxelized point clouds downsampled by fractional factor, using lookup-tables (LUT) constructed from self-similarities their own neighborhoods. The proposed was developed densify and increase the precision of clouds, can be used, for example, as improve compression rendering. geometry, but we also interpolate texture averaging colors adjacent neighbors, completeness. Our technique, understand, is first specifically intra-frame super-resolution arbitrary...

We propose a complete system to enable progressive coding with quality scalability of the mesh geometry, in MPEG's state-of-the-art Video-based Dynamic Mesh Coding (V-DMC) framework. In particular, we an alternative method for encoding subdivision wavelet coefficients V-DMC, using zerotree approach that works directly native 3D space. This allows us identify parent-child relationships amongst across different levels, which can be used achieve efficient and versatile mechanism. demonstrate...

Geometry-based point cloud compression (G-PCC) has been rapidly evolving in the context of international standards. Despite inherent scalability octree-based geometry description, current G-PCC attribute techniques prevent full for compressed clouds. In this paper, we present a solution to add attributes using region-adaptive hierarchical transform (RAHT), enabling reconstruction only portion original bitstream. Without information, one cannot compute weights which RAHT relies on calculate...

We present a method for post-processing point clouds' geometric information by applying previously proposed fractional super-resolution technique to clouds compressed and decoded with MPEG's G-PCC codec. In some sense, this is continuation of that previous work, which requires only down-scaled cloud scaling factor, both are provided the For non-solid clouds, an priori down-scaling required improved efficiency. The compared GPCC itself, as well machine-learning-based techniques. Results show...

<div>We present a method to super-resolve voxelized point clouds down-sampled by fractional factor, using look-up-tables (LUT) constructed from self-similarities its own neighborhoods. Given cloud geometry Vd, and corresponding down-sampling factor s, the proposed determines set of positions that may have generated estimates which these were indeed occupied (super-resolution). Assuming is approximately self-similar at different scales, LUTs relating neighborhood configurations with...

We present a method for post-processing point clouds’ geometric information by applying previously proposed fractional super-resolution technique to clouds compressed and decoded with MPEG’s G-PCC codec. In some sense, this is continuation of that previous work, which requires only downscaled cloud scaling factor, both are provided the For non-solid clouds, an priori down-scaling required improved efficiency. The compared GPCC itself, as well machine-learning-based techniques. Results show...

<p>We propose a complete system to enable progressive coding with quality scalability of the mesh geometry, in MPEG’s state-of-the-art Video-based Dynamic Mesh Coding (V-DMC) framework. In particular, we an alternative method for encoding subdivision wavelet coefficients V-DMC, using mesh-based zerotree approach. The proposed works directly native 3D space. It allows us identify parent-child relationships amongst across different levels, which can be used achieve efficient and...

<p>We propose a complete system to enable progressive coding with quality scalability of the mesh geometry, in MPEG’s state-of-the-art Video-based Dynamic Mesh Coding (V-DMC) framework. In particular, we an alternative method for encoding subdivision wavelet coefficients V-DMC, using mesh-based zerotree approach. The proposed works directly native 3D space. It allows us identify parent-child relationships amongst across different levels, which can be used achieve efficient and...

Geometry-based point cloud compression (G-PCC) has been rapidly evolving in the context of international standards. Despite inherent scalability octree-based geometry description, current G-PCC attribute techniques prevent full for compressed clouds. In this paper, we present a solution to add attributes using region-adaptive hierarchical transform (RAHT), enabling reconstruction only portion original bitstream. Without information, one cannot compute weights which RAHT relies on calculate...

<div>We present a method to super-resolve voxelized point clouds down-sampled by fractional factor, using look-up-tables (LUT) constructed from self-similarities its own neighborhoods. Given cloud geometry Vd, and corresponding down-sampling factor s, the proposed determines set of positions that may have generated estimates which these were indeed occupied (super-resolution). Assuming is approximately self-similar at different scales, LUTs relating neighborhood configurations with...

Geometry-based point cloud compression (G-PCC) has been rapidly evolving in the context of international standards. Despite inherent scalability octree-based geometry description, current G-PCC attribute techniques prevent full for compressed clouds. In this paper, we present a solution to add attributes using region-adaptive hierarchical transform (RAHT), enabling reconstruction only portion original bitstream. Without information, one cannot compute weights which RAHT relies on calculate...

Geometry-based point cloud compression (G-PCC) has been rapidly evolving in the context of international standards. Despite inherent scalability octree-based geometry description, current G-PCC attribute techniques prevent full for compressed clouds. In this paper, we present a solution to add attributes using region-adaptive hierarchical transform (RAHT), enabling reconstruction only portion original bitstream. Without information, one cannot compute weights which RAHT relies on calculate...

In this paper, we evaluate the effects of occlusions in performance a face recognition pipeline that uses ResNet backbone. The classifier was trained on subset CelebA-HQ dataset containing 5,478 images from 307 classes, to achieve top-1 error rate 17.91%. We designed 8 different occlusion masks which were applied input images. This caused significant drop performance: its for each mask became at least two times worse than before. order increase robustness under occlusions, followed...