13. NeuralNet Filter
Up to this point, we have focused on matching whole images. The NeuralNet filter matches a segment of an image(s).
As we have seen, accurate matching via the Neural Filter requires many matching pairs. Preparing matching pairs for whole images means listing all pairs in the match.txt file.
Preparing matching pairs for image segments is much harder; therefore, rather than using the NeuralFilter for image segments, we will use the Unsupervised Filter for image segments. As we have seen, the Unsupervised Matching for image segments is not as accurate as the Neural Filter.
Matching image segments is not the primary focus of the current version of the ImageFinder. If you need accurate Segment Matching, you need customization.
Chapter contents include:
13.1 Trademark Recognition
13.2 Key Segment
13.3 Training
13.4 1:N Matching
13.5 Results
13.6 Mr. Potato
13.7 Monopoly
Copyright (c) 1998 - 2006 Attrasoft, Inc. All rights reserved.