We are working on a computer vision model that is built on excitatory and inhibitory input. It is aimed to discriminate objects that may be parts of other objects. E.g. the letter F and the letter E. In order to quantify the effectiveness of our approach we would like to apply it to an appropriate benchmark data set which contains objects that might be part of other objects.
Can somebody suggest any benchmark data set or perhaps some indoor/outdoor objects which we can use to create our own data set?
Hi,
have a look at http://www.cs.cmu.edu/~cil/v-images.html where you can find a huge amount of benchmark data sets.
You can also check http://www1.cs.columbia.edu/CAVE/databases/ or http://staff.science.uva.nl/~aloi/.
Best, Juan Luis
Dear George, computer vision is a very interesting research field. Nevertheless, our human (non-computerized) physiological vision system works very well, expecially in its capability to adapt itself to a very wide range of light/dark situations. I think that some preliminar readings on either human or animal eye anatomo-physiological behaviour would help you very much. For example: you speak of excitatory-inhibitory input. But, what's about the On-Off behaviour of our retinal three layers neurons?
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing