First I want to say thanks for building this awesome library -- it's a breeze to set up and has awesome performance.
Is it currently possible for users to parameterize the number of units in the feature vector extracted for each image? My intuition is that increasing the size of this vector will create more opportunities for images to match, and will increase the precision of float values that measure the similarity between two images (e.g. will give the dist and score values more information with which to compute distance/similarity values).
First I want to say thanks for building this awesome library -- it's a breeze to set up and has awesome performance.
Is it currently possible for users to parameterize the number of units in the feature vector extracted for each image? My intuition is that increasing the size of this vector will create more opportunities for images to match, and will increase the precision of float values that measure the similarity between two images (e.g. will give the
dist
andscore
values more information with which to compute distance/similarity values).