Present in the meeting: Luis and Tom.
– I showed Tom the results of the clustering of superpixels using a covariance matrix to describe each superpixel
and RGB to describe the individual pixels inside a superpixel. The superpixels where obtained using SLICO and
having an initial number of superpixels of (total_image_pixels / (25*25)).
Comments by Tom:
– Show the probability map.
– If the probability map gives total confidence to the superpixels that actually form part of the tool and less confidence
to the superpixels that are false positives then use hysteresis thresholding. The mechanism is: accept the ones that
we are really confident that are tool, reject those in which we are really confident that are background, and accept
the ones in the middle section of the threshold only if they are neighbours of pixels belonging to tool (upper section).
– Move to supervised.