General Vision Problem

Biological vision systems have evolved over millions of years into efficient and extremely robust entities with a level of perception and understanding that greatly surpasses the creations of modern machine vision. Machine vision has been very successful in finding solutions to specific, well constrained problems such as optical character recognition or fingerprint recognition. In fact machine vision has surpassed human vision in many such closed domain tasks. However, the performance of machine vision systems deteriorates dramatically under non-ideal conditions such as high-contrast illumination or general problem domains such as image retrieval.

It seems that biology provides us with the only existential proof that the general vision problem can be solved. If not for the fact that humans and animals survive in the general environment, proficiently using their vision systems, one would be tempted to conclude that the general vision problem was impossible to unravel. How can a biological or machine system which just captures two dimensional visual information from a view of a cluttered field even attempt to reason with and function in the environment? An accurate detailed spatial model of the environment is difficult to compute and the whole problem of scene analysis is ill-posed.

However biological systems do not really attempt to solve the general vision problem. For example I can't build an accurate spatial world model of the scene I look at ... biological systems have evolved to process visual data to extract just enough information to perform the reasoning for everyday tasks that are part of survival. The hawk is able to detect the spatio-temporal movement of a tiny mouse at a distance; a macaque monkey can locate and dextrously pick a bright red fruit. Visual processing is associated with a task that constrains the reasoning and provides contextual task information that helps make vision possible.

Sumitha Balasuriya