README

MetroVisionML:
	A supervised machine learning classification system built on top of opencv.
	The purpose of this program is to be able to create a system that can train a classifier
	given images placed in a database with image and classification information.
	Once trained the software provides an interface to classify objects using the given model.	

	Files:
		db/db.sql : The sql to generate a database in the format the program expects
		db/metrovisionML.py : a python program that will store the files that will be uploaded into the
			database using a folder structure in which images are stored in folders named after
			the classification information.
			You will need the YUV range information of the classification if the classification name isn't
			in there yet.
			BE CAREFUL! If it doesn't ask you to add yuv-information it means there exists that classification
			already in the database. orangeT for the aibo should not be the same orangeT for the Scribbler!
			Instead, best to give it a name like orangeT-aibo-6/10/2010 or something unique!
		db/convertSet.rb : a ruby program that will convert an openCv featureSet file into a csv file
			usage: convertSet.rb [openCv input filename] [csv output filename]
		db/convertResp.rb : a ruby program that will convert an openCv responses file into a csv file
			usage: convertResp.rb [openCv input filename] [csv output filename] [# of objects classified]
		models/surveryor-o-l.yaml: a model to classify surveyor images between negatives, orangeT's, 
			and greenLs

Dependencies:
	mysqlpp: http://tangentsoft.net/mysql++/
	yaml-cpp: http://code.google.com/p/yaml-cpp/
	opencv: http://opencv.willowgarage.com/wiki/Welcome
	player: http://playerstage.sourceforge.net/	

Once dependencies are installed, just run make
		you may need to change the MYSQL_ definitions to point to where mysql++ is installed in the Makefile

Programs that will be installed:
	bin/
		AngleTester: Tests the dataset using a combination histogram of angles, 
		             histogram of edge length, average color
		CalSaver: Writes as files into the folder that called it the files in a 
                          dataset and their YUV counterparts for use in the Cal program
		DBDatasetTest: Will iterate over all the files in a dataset and 
			       display them onscreen, press any button to iterate
		FileTester: Uses the featureset and the response set as files to do 
			    classification and testing
		VisualCortexTest: Given a model and an image, the system will output a classification
		eyes: This program will take frames from a accessible webcam and perform classifcation 
		      using the given model
		PlayerClassify: Given a model this will display images from player stage (it converts from jpeg,
				so it works with the surveyor now, need to check for JPEG or RAW, and convert later
				to see if it works on the AIBO) and if you hit c it will save the image,
				if you hit w it will try to classify the image, printed on the terminal.

See TUTORIAL and TODO for more information