Computer Vision Archives - Biomedical Computer Vision

Embodied AI

Santiago — Fri, 05 Apr 2024 20:59:35 +0000

The post Embodied AI appeared first on Biomedical Computer Vision.

Robustness and Interpretability

Santiago — Fri, 15 Oct 2021 14:24:14 +0000

Computer Vision systems have achieved remarkable performances across a wide variety of tasks, such as recognition, segmentation, detection, and generation. However, these systems have also been shown to be vulnerable against semantically-meaningless perturbations. In particular, recent works have shown that these systems, while accurate, lack robustness. This property is undesirable for intelligent systems on which we wish to rely on in the real world. In the Center, we have worked on robustness on various dimensions. In particular, we have (1) designed biologically-inspired techniques to improve robustness, (2) proposed novel semantically-oriented dimensions for the assessment of the robustness, (3) studied how inexpensive techniques during system deployment can provide robustness benefits, (4) investigated the pervasiveness of the lack of robustness in the medical domain, and (5) shown how techniques for improving robustness can be harnessed to improve the performance of super-resolution systems.

The post Robustness and Interpretability appeared first on Biomedical Computer Vision.

Facial Expression Understanding

Cerbero — Thu, 08 Jul 2021 02:54:16 +0000

Human facial expression interpretation has been a classic field of study in psychology, and it has benefited from seminal contributions by renowned researchers such as P. Ekman, who characterized and studied the manifestation of prototypical emotions through changes in facial features. From the computer vision perspective, solving the problem of automated facial expression interpretation is a cornerstone towards high-level human computer interaction, and its study has become an active topic of research in the last decades.

In order to study facial expressions in a systematic way, Ekman and his collaborators designed the Facial Action Coding System (FACS). FACS relies on identifying visible local appearance variations in the human face, called Action Units (AUs), produced by to contraction or relaxation in any of its 30 muscles (e.g., a raised eyebrow). AUs constitute therefore a natural physiological basis for face analysis, in which any facial expression can be, potentially, represented by their combinations.

The existence of a physiological basis for a computer vision domain is a rare luxury, as it allows focusing on the essential atoms of the problem and, by virtue of their multiple possible combinations, opens the door to a wide range of applications beyond the emotion classification domain such as psychological, medical, legal, entertainment, etc.

Figure 1. Emotions and its action units

The post Facial Expression Understanding appeared first on Biomedical Computer Vision.

RGB-D Scene Understanding

Cerbero — Thu, 08 Jul 2021 02:47:03 +0000

The post RGB-D Scene Understanding appeared first on Biomedical Computer Vision.

3D Vision

Cerbero — Thu, 03 Jun 2021 03:23:28 +0000

Encouraged by evolving research fields such as Artificial Reality, Autonomous vehicles, and scene understanding, 3D Vision problems have gained interest recently. Many of the tasks studied in 3D Vision are inspired by its 2D counterpart. However, the extending of deep learning into depth and 3D information can unlock a variety of applications. Contrary to 2D Vision problems that study 2D standard-images, 3D data is structured in different formats, including RGB-D images, voxel grids, point clouds, and meshes. This diversity and the need for low-computational-cost processing is a challenge for research in this area.

The post 3D Vision appeared first on Biomedical Computer Vision.