English
Grupo de Processamento de Sinal Multimédia
Mais Informação
+351 217650590
The Multimedia Signal Processing Group of IT-IUL (MSP-IUL) has the objective of developing technology and advancing the state-of-the-art in the areas of visual information analysis, representation, processing and quality of experience evaluation, mostly relying on cutting edge deep learning approaches.
In terms of visual information analysis, the main objective of the group’s work is to automatically extract knowledge from visual content to help make more informed decisions. In the area of computer vision this includes, for example, the development of advanced techniques for estimating depth maps from visual scenes, semantic object detection, and scene understanding. In the area of immersive multimedia applications this includes, for example, developing techniques for enhanced post-production and scene enhancement in AR/VR environments.
In terms of visual information representation, the MSP-IUL group works on the development of coding techniques for advanced visual modalities, by focusing on signals generated by emerging event-based cameras, as well as higher-dimensional visual modalities, like light fields and point clouds.
In terms of visual information processing, the MSP-IUL group works on the development of more effective processing techniques for traditional, as well as higher-dimensional, visual modalities. Examples of possible processing techniques include (spatial, angular and temporal) super-resolution, view synthesis and style transfer.
In terms of quality of experience evaluation, the MSP-IUL group works on the development of subjective evaluation methodologies and objective visual quality assessment metrics. Since the focus is mainly on higher-dimensional visual modalities, in addition to the usual problems, scene navigation problems are also considered. The quality of experience related to visual signals generated from event-based cameras is also being considered.
The experimental laboratory of the MSP-IUL group is currently equipped with various types of imaging acquisition devices, such as industrial 2D color cameras, light field cameras, a 3D scanner and a stereo event-based camera setup. In terms of computation capabilities, the MSP-IUL lab is equipped with multiple servers with NVIDIA RTX 3080, 3090 and 4090 GPUs and has access to the Iscte datacenter with a Computing Cluster of 12 nodes equipped with two Intel Xeon CPUs each and 8 GPU Accelerated Computing nodes for AI equipped with one NVIDIA Ampere A100 GPU each.
The MSP-IUL is regularly offering positions to highly motivated PhD students.
