RESEARCH PAPER / MILCOM 2022 / Nov 2022
The continued demand for spectrum has pressured governments to make more spectrum available to the commercial sector. This in turn has put pressure on government users of the spectrum to be more efficient even as their own demand for data grows. As a result, the US government is considering using...
RESEARCH PAPER / Workshop at NIPS 2022 - Memory in Artificial and Real Intelligence / Nov 2022
The Predicting Media Memorability task in the MediaEval evaluation campaign has1 been running annually since 2018 and several different tasks and data sets have been2 used over those years. This has allowed us to compare the performance of many3 techniques on the same data and in a reproducible way and...
RESEARCH PAPER / Web3D / Nov 2022
With new use cases in markets such as gaming, IoT, PC, and extended reality (AR/VR/MR), there is an increasing demand for richer haptic experiences. High-definition (HD) haptics, with effects ranging from subtle to sharp, textured effects that simulate different surfaces and sensations, and increasingly efficient vibration motors, are becoming the...
RESEARCH PAPER / IEEE Access / Oct 2022
Radio access network (RAN) technologies continue to witness an insatiable rate of development progress, with Open-RAN gaining the most recent momentum. In the O-RAN structure, the RAN intelligent controller (RIC) provides a host for various AI/ML models. This article introduces principles of machine learning (ML), in particular, reinforcement learning (RL)...
RESEARCH PAPER / MOBICOM 22 - Winter / Oct 2022
With the continuous growth of the Internet of Things (IoT),the trend of increasing connection to the Internet of billionsof new IoT devices will continue. To increase network capa-bility to support a large number of active devices accessing anetwork (i.e.,massive IoT connectivity), this work presentsIoT-ResQ, a warm-started quantum annealing-based multi-device detector...
RESEARCH PAPER / IEEE International Conference in Image Processing / Oct 2022
Generative adversarial networks (GANs) have proven to be surprisingly efficient for image editing by inverting and manipulating the latent code corresponding to an input real image. This editing property emerges from the disentangled nature of the latent space. In this paper, we identify that the facial attribute disentanglement is not...
RESEARCH PAPER / European Conference on Computer Vision / Oct 2022
We present a new encoder architecture for GAN inversion. The task is to reconstruct a real image from the latent space of a pre-trained Generative Adversarial Network (GAN). Unlike previous encoder-based methods which predict only a latent code from a real image, the proposed encoder maps the given image to...
RESEARCH PAPER / Breizh Video Tech / Oct 2022
The increasing popularity of virtual, augmented, and mixed reality (VR/AR/MR) applications is driving the media industry to explore the creation and delivery of new immersive experiences. A volumetric video consists of a sequence of frames, where each frame is a static three-dimensional (3D) representation of a real-world object or scene...
RESEARCH PAPER / ACM Multimedia 2022 Workshop / Oct 2022
We propose in this paper a new paradigm for facial video compression. We leverage the generative capacity of GANs such as StyleGAN to represent and compress a video, including intra and inter compression. Each frame is inverted in the latent space of StyleGAN, from which the optimal compression is learned....
RESEARCH PAPER / ACM Multimedia 2022 Workshop / Oct 2022
Point cloud compression (PCC) serves as a crucial phase in various 3-D applications, owing to the universality of the point cloud format. Ideally, 3D point clouds endeavor to depict object/scene surfaces that are continuous. Practically, as a set of discrete samples, point clouds are locally disconnected and sparsely distributed. This...
RESEARCH PAPER / ESANN 2022 / Oct 2022
Convolutional neural networks (CNN) are often computationally demanding for mobile devices. Offloading some computation lowers this burden: initial convolutional layers are processed on a smartphone, the resulting high dimensional features transmitted, and latter layers processed in the cloud/edge/another device. To improve this process, we propose Dynamic Switch, a convolutional subnetwork...
RESEARCH PAPER / Elsevier book : Immersive Video Technologies / Oct 2022
This chapter presents the state-of-the-art in the coding (or compression) of dynamic 3D mesh models. Section 1.1 introduces the motivations behind using 3D meshes for volumetric video. In Section 1.2, some fundamental mesh concepts are explained, which are required in order to understand the state-of-the-art that follows. Section 1.3 briefly...
RESEARCH PAPER / VTC 2022 / Sep 2022
A reconfigurable intelligent surface (RIS) can be used to control the propagation of electromagnetic waves (EM). Deployment of RIS units in radio environment allows to steer the transmitted EM waves to areas that are otherwise shadowed by buildings or geographic formations such as hills. Since the RIS is a passive...
RESEARCH PAPER / VTC 2022 / Sep 2022
Reconfigurable Intelligent Surface (RIS) consists of mostly-passive elements capable of electronically steering the impinging signal by configuring the phase shifts. However, achieving infinite phase resolution is infeasible and needs to be quantized for practical implementation. In this paper, we propose an unsupervised learning-based method to estimate the optimal discrete RIS...
RESEARCH PAPER / IEEE Transactions on Multimedia / Sep 2022
Geometric data acquired from real-world scenes, e.g., 2D depth images, 3D point clouds, and 4D dynamic point clouds, have found a wide range of applications including immersive telepresence, autonomous driving, surveillance, etc. Due to irregular sampling patterns of most geometric data, traditional image/video processing methodologies are limited, while Graph Signal...
RESEARCH PAPER / IEEE Wireless Communication Letters / Sep 2022
Hybrid beamforming provides a cost effective strategy towards practical deployment of massive multiple-input multiple-output (MIMO) systems. Since the hybrid precoder-combiner evaluation requires channel state information, the computation is performed at the receiver and the evaluated precoders are communicated back to the transmitter. This transmission overhead associated with the precoder feedback...
RESEARCH PAPER / SMPTE Media Technology Summit 2022 / Sep 2022
Where content producers had to make drastic choices when allocating the limited contrast and colors in SDR, HDR offers them the possibility to show more - meaning telling stories with greater freedom and flexibility. However, while future-proof, HDR content presents a challenge of its own: The wide variation in capabilities...
RESEARCH PAPER / IEEE Globecom 2022 / Sep 2022
Among the new technologies under research for Beyond-5G and 6G wireless communications, THz and sub-THz frequencies are receiving considerable attention thanks to their huge potentially available bandwidth. Phase noise is one of the most critical impairments in these bands because of its much higher impact on demodulation performance compared to...
RESEARCH PAPER / ACM SCA (Symposium on Computer Animation) / Sep 2022
Human motion synthesis and editing are essential to many applications like video games, virtual reality and film post-production. However, they often introduce artefacts in motion capture data, which can be detrimental to the perceived realism. In particular, footskating is a frequent and disturbing artefact, which requires knowledge of foot contacts...
RESEARCH PAPER / EUVIP 2022 / Sep 2022
This paper presents a solution to the dynamic mesh compression Call for Proposals (CfP) that was recently launched by the MPEG 3D Graphics Coding (MPEG-3DGC) group. The proposed method begins with a per-frame mesh decimation using QSlim based on quadric error metrics. The decimated mesh geometry and topology are then...