Custom Search

Sunday, November 21, 2010

Why Vision?

Among those hottest words in Computer Science topic of VideoLectures, Machine learning is the king without any doubt as it is almost the basic of CS nowadays, for every research area. Then the hot words are Semantic Net, Data Mining, Computer Vision, Natural Language Processing... I think we can combine Data Mining and Machine Learning into one "extended" research area and combine Semantic Net, Computer Vision, Natural Language Processing and even Image Processing into another "extended" research area as modern Vision and Linguistic research are always related with Semantic Analysis. Nevertheless, applying ML technologies on Vision and Linguistic research dominates the main trend.

Here comes my puzzle, which haunting me for a long time: Why so many people focus on Vision study? Vision is just a part of our sensation system and there are other crucial senses, such as odor system, people pay much more less attention on them. Why vision?

Christof Koch explained why he focuses on Vision in his famous book THE QUEST OF CONSCIOUSNESS as follows:

First, humans are visual creatures. This is reflected in the large amount of brain tissue dedicated to the analysis of images, and in the importance of seeing in daily life. If you have a cold, for instance, your nose becomes stuffy and you may lose your sense of smell, but this impairs you only mildly. A transient loss of vision, as occurs during snow blindness on the other hand, devastates you.

Second, visual percepts are vivid and rich in information. Pictures and movies are highly structured, yet easy to manipulate using computer-generated graphics.

Third, as noted already by the young philosopher Arthur Schopenhauer in 1813, vision is more easily deceived than any of the other senses. Blablabla...

Last and  most important, the neuronal basis of many visual phenomena and illusions has been investigated throughout the animal kingdom. Perceptual neuroscience has advanced to such a point that reasonably sophisticated computational models have been constructed and have proven their worth in guiding experimental agendas and summarizing the data.

3 comments:

  1. hi, i see you graduated from Zhejiang University. I am also just graduated from Zhejiang University this July. Now I am in the United States and I also plan to study Machine Learning. My major was pure mathematics and now i am in the statistics department. Nice to meet you here.

    ReplyDelete
  2. @ Honglang: Nice to meet you too! I am not a statistic guy but we learn a lot from statistics department.

    ReplyDelete
  3. hi there i guess you know who i am. i'll be a subscriber and commenter here from now on

    well in my opinion, first, vision is interesting, which is apparent. second, like you quoted from Chris, vision is the major source of info that we receive everyday, other sensors do help but don't really dominate like vision. another potential reason is that it might not be a very demanding field; by demanding i mean the need for fancy equipments and heavy-duty experiments. with a bunch of images and videos you're off and running!

    ReplyDelete

Custom Search