This repository contains code and examples that demonstrate the ability for artificial neural networks to classify images of insect specimens. The layout is as follows:
- lib: contains a utility class that does image pre-processing
- data/img: contains training and testing data with pictures of beetles and butterflies
- data/traindata: contains tab-separated tables of "fingerprints" of training images
- data/ai: contains a stored AI
- script/splitter.pl: a naive implementation of an image segmentation algorithm
- script/traindata.pl: generates "fingerprints" of images as tab-separated tables
- script/trainai.pl: feeds training data into the ANN
- script/classify.pl: classifies a directory of out-of-sample images
- ImageMagick (c library) and Image::Magick (perl bindings)
- FANN (c library) and AI::FANN (perl bindings)
- Bio::Phylo (for logging)