http://hdl.handle.net/10071/5707 2026-05-25T23:53:01Z http://hdl.handle.net/10071/37275 Título próprio: LFVS-Mamba: State-space model for light field view synthesis Autoria: Zubair, M.; Nunes, P.; Conti, C.; Soares, L. D. Resumo: Light Field View Synthesis (LFVS) methods using Convolutional Neural Networks (CNNs) and Vision Transformers (VTs) have been extensively studied: CNNs excel at learning local spatial features via hierarchical receptive fields but cannot capture long-range global dependencies, while VTs inherently model global context through self-attention at the cost of quadratic computation and memory complexity. To address these issues, we propose LFVS-Mamba, which integrates a State-Space Module (SSM) with a Selective Scanning Mechanism to efficiently capture long-range dependencies. LFVS-Mamba processes 2D slices of the 4D LF to fully exploit spatial context, complementary angular information, and depth cues. The LFVS-Mamba comprises three modules to progressively synthesize dense LFs: (i) Shallow Feature Extraction (SFE), (ii) Spatial-Angular Depth Feature Extraction (SADFE), and (iii) Angular Upsampling (AU). Experimental results on standard LF benchmarks demonstrate that LFVS-Mamba consistently outperforms existing methods. 2025-01-01T00:00:00Z http://hdl.handle.net/10071/36082 Título próprio: Learning-based lossless event data compression Autoria: Sezavar, A.; Brites, C.; Ascenso, J. Resumo: Emerging event cameras acquire visual information by detecting time domain brightness changes asynchronously at the pixel level and, unlike conventional cameras, are able to provide high temporal resolution, very high dynamic range, low latency, and low power consumption. Considering the huge amount of data involved, efficient compression solutions are very much needed. In this context, this paper presents a novel deep-learning-based lossless event data compression scheme based on octree partitioning and a learned hyperprior model. The proposed method arranges the event stream as a 3D volume and employs an octree structure for adaptive partitioning. A deep neural network-based entropy model, using a hyperprior, is then applied. Experimental results demonstrate that the proposed method outperforms traditional lossless data compression techniques in terms of compression ratio and bits per event. 2024-01-01T00:00:00Z http://hdl.handle.net/10071/35443 Título próprio: Best practices for accurate results using numerical solvers for microwave body screening Autoria: Martins, R. A.; Godinho, D.; Felício, J. M.; Savazzi, M.; Costa, J. R.; Conceição, R.; Fernandes, C. A. Resumo: In this paper, we indicate best practices that should be observed when using numerical solvers for microwave body sensing. We show the impact of not minding these aspects in the case of microwave breast scanning, using the Computer Simulation Technology software tool. To this end we simulate a homogeneous breast with a 5-mm radius spherical tumor placed inside. The breast is illuminated by a broadband antenna that operates in the 2-6 GHz band. The scattering parameters are then processed to reconstruct the reflectivity map of the breast. The results highlight that the conclusions drawn from simulations may be misleading or meaningless when the solver type or positioning of model elements (body and antennas) are not carefully applied. This is particularly critical when considering more complex scenarios, such as inhomogeneous or multilayer body models. 2024-01-01T00:00:00Z http://hdl.handle.net/10071/35432 Título próprio: Antenna position layout and frequency impact on tumor detection in microwave breast imaging Autoria: Martins, R. A.; Felício, J.; Costa, J. R.; Fernandes, C. A. Resumo: We present a systematic study in which we assess the antenna position layout and frequency point distribution of a MWI system, that can potentially improve tumor detection and minimize acquisition time. To this end, we performed measurements on a dry MW setup, using a slot-based antenna in the [2]–[5] GHz frequency range to scan an anthropomorphic breast phantom, with two different tumor positions, for 40 angular positions. Imaging and tumor-to-clutter ratio metric showed that there is a specific number of angular positions and frequency points beyond which the quality of imaging results does not increase substantially. We found the optimal frequency band for this kind of setup and that the use of lower frequencies seems more beneficial than the use of higher ones. Moreover, distributions of antenna position other from the regular circular one, should be explored further since it showed a decrease of imaging artefacts. 2024-01-01T00:00:00Z