Help on htm vision problem

[afee]

Hi all,

I am interested in using HTM for invariant object recognition. I read the theory of both generations of HTM algorithm(zeta1 and CLA). I think zeta1 is more suitable for rotation/pose invariant object recognition. But, nupic 1.7.1 is not available at http://www.numenta.com/legacysoftware.php now. Is it possible to use CLA for this purpose?

[rhyolight]

@afee

Is it possible to use CLA for this purpose?

You can try, we have a repo with our vision code at GitHub - numenta/nupic.vision: Tools for using Numenta/nupic on visual problems like image recognition.. But this is not our current focus. Be sure to search the forums for more info. Here is a good topic:

MNIST test problem

Hi, I am working now on image_test at nupic.research/projects for MNIST classification. By running experiment: python experiment.py I have got the following error: ERR: Matching Python module for ImageSensor not found. [/home/binh/nupic/nupic.core/src/nupic/engine/RegionImplFactory.cpp line 431] Could you pls tell me, where can I import the module ImageSensor? Best thanks, Binh