Early Segmentation Results

Posted: February 29, 2016 at 8:47 am

The following image is a crop from the segmented panorama previously posted. The computer is still extracting pixels from the original image using these segmented regions, where the current count exceeds 50,000.


The image is so large that it exceed available GPU memory. The program divides it into chunks that will fit in GPU memory, segment those chunks, and then recombine them into a single image in main memory. One issue with this is that the seams between chunks are likely to intersect segmented regions, so I need to do some manual merging of regions divided in this process.

Once these regions have been extracted then the first thing I’ll do is use the collage code written for Watching and Dreaming to get a sense of how these percepts look in a single composition. After that I can revisit the Self-Organized Map and go from there.