A5002481506- Remote Sensing and LiDAR Applications
- 3D Shape Modeling and Analysis
- 3D Surveying and Cultural Heritage
- Computer Graphics and Visualization Techniques
- Music and Audio Processing
- Speech and Audio Processing
- Speech Recognition and Synthesis
- Image and Signal Denoising Methods
- Advanced Data Compression Techniques
- Advanced Malware Detection Techniques
- Gait Recognition and Analysis
- Hand Gesture Recognition Systems
- Image and Object Detection Techniques
- Internet Traffic Analysis and Secure E-voting
- Landslides and related hazards
- Image Processing and 3D Reconstruction
- Advanced Image Processing Techniques
- Image Processing Techniques and Applications
- Advanced Steganography and Watermarking Techniques
- Robotics and Sensor-Based Localization
University of Padua
2021-2024
In the current age, users consume multimedia content in very heterogeneous scenarios terms of network, hardware, and display capabilities. A naive solution to this problem is encode multiple independent streams, each covering a different possible requirement for clients, with an obvious negative impact both storage computational requirements. These drawbacks can be avoided by using codecs that enable scalability, i.e., ability generate progressive bitstream, containing base layer followed...
The recent integration of generative neural strategies and audio processing techniques have fostered the widespread synthetic speech synthesis or transformation algorithms. This capability proves to be harmful in many legal informative processes (news, biometric authentication, evidence courts, etc.). Thus, development efficient detection algorithms is both crucial challenging due heterogeneity forgery techniques.This work investigates discriminative role silenced parts shows how first digit...
Many recent cloud or edge computing strategies for automotive applications require transmitting huge amounts of Light Detection and Ranging (LiDAR) data from terminals to centralized processing units. As a matter fact, the development effective Point Cloud (PC) compression that preserve semantic information, which is critical scene understanding, proves be crucial. Segmentation have always been treated as two independent tasks; however, since not all classes are equally important end task,...
Learned image compression codecs have recently achieved impressive performances surpassing the most efficient coding architectures. However, approaches are trained to minimize rate and distortion which often leads unsatisfactory visual results at low bitrates since perceptual metrics not taken into account. In this paper, we show that conditional diffusion models can lead promising in generative task when used as a decoder, that, given compressed representation, they allow creating new...
The widespread usage of point clouds (PC) for immersive visual applications has resulted in the use very heterogeneous receiving conditions and devices, notably terms network, hardware, display capabilities. In this scenario, quality scalability, i.e., ability to reconstruct a signal at different qualities by progressively decoding single bitstream, is major requirement that yet be conveniently addressed, most learning-based PC coding solutions. This paper proposes scalability scheme, named...
In recent years the advancements in neural networks field have fostered advent of end-to-end learned coding schemes capable efficient image representations that reduce required storage space and transmission time.In general, features produced by these encoders are entropy-efficient permit reconstructing coded with low distortion. However, whenever they applied to a generic image, its latent representation might not be optimal one feature since network parameters were trained generalize on...
Nowadays living environments are characterized by networks of inter-connected sensing devices that accomplish different tasks, e.g., video surveillance an environment a network CCTV cameras. A malicious user could gather sensitive details on people's activities eavesdropping the exchanged data packets. To overcome this problem, streams protected encryption systems, but even secured channels may still leak some information. In paper, we show it is possible to infer visual intercepting...
Recent advances in deep learning and computer vision have made the synthesis counterfeiting of multimedia content more accessible than ever, leading to possible threats dangers from malicious users. In audio field, we are witnessing growth speech deepfake generation techniques, which solicit development synthetic detection algorithms counter mischievous uses such as frauds or identity thefts. this paper, consider three different feature sets proposed literature for task present a model that...