Deep Hierarchies in the Primate Visual Cortex: What Can We Learn for Computer Vision?

Abstract

Abstract-Computational modeling of the primate visual system yields insights of potential relevance to some of the challenges that computer vision is facing, such as object recognition and categorization, motion detection and activity recognition or vision-based navigation and manipulation. This article reviews some functional principles and structures that are generally thought to underlie the primate visual cortex, and attempts to extract biological principles that could further advance computer vision research. Organized for a computer vision audience, we present functional principles of the processing hierarchies present in the primate visual system considering recent discoveries in neurophysiology. The hierarchal processing in the primate visual system is characterized by a sequence of different levels of processing (in the order of ten) that constitute a deep hierarchy in contrast to the flat vision architectures predominantly used in today's mainstream computer vision. We hope that the functional description of the deep hierarchies realized in the primate visual system provides valuable insights for the design of computer vision algorithms, fostering increasingly productive interaction between biological and computer vision research.

Using surfaces and surface relations in an Early Cognitive Vision system

“…There is evidence that also in the human visual system, there exist quite an amount of genetic precoding. However, there is also clear evidence for learning, even in rather early areas such as V1 and V2 (see, e.g., [45]). In our future work, we intend to relax some of the hard-wired assumptions by learning.…”

“…There exists overwhelming evidence that the human visual system computes a large variety of aspects in the two large areas V1 and V2 in parallel, covering both edge and surface aspect (see, e.g., [45]). That means that at least for local low-level feature processing on the level of V1 and V2, no major selection process seems to take place for reducing computational complexity.…”

Image Quality Assessment Based on Multi-Order Local Features Description, Modeling and Quantification

“…Besides, the proposed method has a comparable performance with CMSD on CSIQ database. In accordance with neurons in the early visual areas, local feature description extracts image features over small local regions [50]. The first-order operator that extracts edges information [51] and high-order operator conveying more knowledge of texture regions [52] are combined to generate more sufficient accurate quality-aware features.…”

Biological Models for Active Vision: Towards a Unified Architecture

“…Attempts to build a comprehensive biologically plausible system have focussed on dynamic field-based models of different aspects of vision [26] and cognitive robots [25]. A good comparative summary of computer vision and biological vision is given in [27]. Figure 1 gives an overview of our system, which is a simplified model of the mammalian brain (for an excellent overview of different visual processing pathways we refer to [28]).…”