def isProperDetector(self,
configString,
detectorData=None,
detectorProperties=None):
'''Try to create or use the specified detector.
Create a test runset and call the detector with it.
Check for proper handling of intentional problems.
'''
detector = easy.getDetector(configString)
if not detector:
print("cannot find or connect to detector")
return False
# make a runset including erroneous files
rs = cvac.RunSet()
file_normal = easy.getLabelable("testImg/italia.jpg")
file_notexist = easy.getLabelable("testImg/doesnotexist.jpg")
file_notconvertible = easy.getLabelable("testImg/notconvertible.xxx")
easy.addToRunSet(rs, file_normal,
cvac.Purpose(cvac.PurposeType.UNPURPOSED, -1))
easy.addToRunSet(rs, file_notexist,
cvac.Purpose(cvac.PurposeType.UNPURPOSED, -1))
easy.addToRunSet(rs, file_notconvertible,
cvac.Purpose(cvac.PurposeType.UNPURPOSED, -1))
results = easy.detect(detector, detectorData, rs, detectorProperties)
# check for number of results: must be the same as runset length
nRunset = 0
for pur in rs.purposedLists:
nRunset += len(pur.labeledArtifacts)
if len(results) != nRunset:
print("incorrect result set size")
return False
# check that the doesnotexist file caused hasLabel==False
found = False
for lbl in results:
# for an erroneous file
if lbl.original.sub == img_notexist.sub or lbl.original.sub == img_notconvertible.sub:
for foundlbl in lbl.foundLabels:
if foundlbl.lab.hasLabel:
print(
"Incorrectly assigned label for an erroneous file."
)
return False
else: #for a normal file
for foundlbl in lbl.foundLabels:
# confidence 0..1
if not 0.0 <= foundlbl.confidence and foundlbl.confidence <= 1.0:
print("Label confidence out of bounds ({0}).".format(
foundlbl.confidence))
return False
# check that either this was not assigned a label, or the
# assigned label is of proper syntax
if foundlbl.lab.hasLabel:
if not isinstance(foundlbl.lab.name, str):
print("Label name must be of string type.")
return False
return True
def isProperDetector( self, configString, detectorData=None, detectorProperties=None ):
'''Try to create or use the specified detector.
Create a test runset and call the detector with it.
Check for proper handling of intentional problems.
'''
detector = easy.getDetector( configString )
if not detector:
print("cannot find or connect to detector")
return False
# make a runset including erroneous files
rs = cvac.RunSet()
file_normal = easy.getLabelable( "testImg/italia.jpg" )
file_notexist = easy.getLabelable( "testImg/doesnotexist.jpg" )
file_notconvertible = easy.getLabelable( "testImg/notconvertible.xxx")
easy.addToRunSet( rs, file_normal, cvac.Purpose( cvac.PurposeType.UNPURPOSED, -1 ))
easy.addToRunSet( rs, file_notexist, cvac.Purpose( cvac.PurposeType.UNPURPOSED, -1 ))
easy.addToRunSet( rs, file_notconvertible, cvac.Purpose( cvac.PurposeType.UNPURPOSED, -1 ) )
results = easy.detect( detector, detectorData, rs, detectorProperties )
# check for number of results: must be the same as runset length
nRunset = 0
for pur in rs.purposedLists:
nRunset += len(pur.labeledArtifacts)
if len(results)!=nRunset:
print("incorrect result set size")
return False
# check that the doesnotexist file caused hasLabel==False
found=False
for lbl in results:
# for an erroneous file
if lbl.original.sub==img_notexist.sub or lbl.original.sub==img_notconvertible.sub:
for foundlbl in lbl.foundLabels:
if foundlbl.lab.hasLabel:
print("Incorrectly assigned label for an erroneous file.")
return False
else: #for a normal file
for foundlbl in lbl.foundLabels:
# confidence 0..1
if not 0.0<=foundlbl.confidence and foundlbl.confidence<=1.0:
print("Label confidence out of bounds ({0}).".format(foundlbl.confidence))
return False
# check that either this was not assigned a label, or the
# assigned label is of proper syntax
if foundlbl.lab.hasLabel:
if not isinstance(foundlbl.lab.name, str):
print("Label name must be of string type.")
return False
return True
def evaluate( contender, runset, printVerbose=False ):
if type(runset) is dict and not runset['runset'] is None\
and isinstance(runset['runset'], cvac.RunSet):
runset = runset['runset']
if not runset or not isinstance(runset, cvac.RunSet) or not runset.purposedLists:
raise RuntimeError("no proper runset")
evalset = runset
print( "---- evaluation:" )
easy.printRunSetInfo( evalset, printArtifacts=printVerbose )
detector = contender.getDetector()
detections = easy.detect( detector, contender.detectorData, evalset,
detectorProperties=contender.detectorProps )
ct = getConfusionTable( detections, origSet=evalset, foundMap=contender.foundMap )
# create result structure
r = EvaluationResult( 0, ct[0], nores=ct[1],
detail=None, name=contender.name )
return r
def test_cvacdatadir(self):
print("testing cvac data dir")
if sys.platform == 'win32':
datadir = os.getenv("CVAC_DATADIR", None)
datadir = datadir.replace("/", "\\")
easy.CVAC_DataDir = datadir
print("Testing using back slashes")
print("Using CVAC_DATADIR as " + datadir)
testset = []
easy.misc.searchDir(testset,
datadir + '\\testImg',
recursive=True,
video=False,
image=True)
runset = cvac.RunSet()
easy.addToRunSet(runset, testset, 'pos')
modelfile = 'detectors/bowUSKOCA.zip'
detector = easy.getDetector("BOW_Detector")
props = easy.getDetectorProperties(detector)
props.verbosity = 3
results = easy.detect(detector,
modelfile,
runset,
detectorProperties=props)
easy.printResults(results)
else:
print("Skipping back slash test on this platform")
#run it again with all forward slashes
datadir = os.getenv("CVAC_DATADIR", None)
datadir = datadir.replace("\\", "/")
print("Testing using all forward slashes")
print("Using CVAC_DATADIR as " + datadir)
easy.CVAC_DataDir = datadir
testset = []
easy.misc.searchDir(testset,
datadir + '/testImg',
recursive=True,
video=False,
image=True)
runset = cvac.RunSet()
easy.addToRunSet(runset, testset, 'pos')
modelfile = 'detectors/bowUSKOCA.zip'
detector = easy.getDetector("BOW_Detector")
props = easy.getDetectorProperties(detector)
props.verbosity = 3
results = easy.detect(detector,
modelfile,
runset,
detectorProperties=props)
easy.printResults(results)
#run it for forward slashes and relative path
origDir = datadir
curDir = os.getcwd()
curDir = curDir.replace("\\", "/")
datadir = datadir.replace("\\", "/")
print("using relative paths for " + curDir + " in data dir " +
datadir)
if datadir.startswith(curDir):
datadir = datadir[len(curDir) + 1:]
easy.CVAC_DataDir = datadir
print("Using CVAC_DataDir as " + datadir)
testset = []
easy.misc.searchDir(testset,
origDir + '/testImg',
recursive=True,
video=False,
image=True)
runset = cvac.RunSet()
easy.addToRunSet(runset, testset, 'pos')
modelfile = origDir + '/detectors/bowUSKOCA.zip'
detector = easy.getDetector("BOW_Detector")
props = easy.getDetectorProperties(detector)
props.verbosity = 3
results = easy.detect(detector,
modelfile,
runset,
detectorProperties=props)
easy.printResults(results)
else:
RuntimeError("Bad datadir")
def doDetect():
easy.detect( detector, modelfile, runset, callbackRecv = callbackRecv, detectorProperties = props)
'''
import easy
# obtain a reference to a Bag of Words (BOW) detector
detector = easy.getDetector( "BOW_Detector" )
# a model for distinguishing Canadian, Korean, and US flags,
# trained previously with a BOW-specific trainer and stored in a file
modelfile = "detectors/bowUSKOCA.zip"
# a test image; the location is relative to the "CVAC.DataDir"
imgfile = "testImg/TestKrFlag.jpg"
# apply the detector type, using the model and testing the imgfile
results = easy.detect( detector, modelfile, imgfile )
# you can print the results with Easy!'s pretty-printer;
# we will explain the meaning of the "unlabeled" and the number
# of the found label later.
# or you can always print variables directly (uncomment the next line):
# print("{0}".format( results ))
print("------- Bag of Words results for flags: -------")
easy.printResults( results )
# let's try a different model, pre-trained not for
# flags but for various corporate logos
print("------- Bag of Words results for corporate logos: -------")
modelfile = "detectors/bowCorporateLogoModel.zip"
imgfile = "corporate_logos/shell/shell2.png"
results = easy.detect( detector, modelfile, imgfile )
rocZip = easy.makeROCdata(rocData_optimal)
'''
Extract only optimal ROC points
'''
detector = easy.getDetector( strDetector )
detectorProps = easy.getDetectorProperties(detector)
#specify either a desired falseAlarmRate or recall, but not both:
priority = "recall"
if priority=="falseAlarmRate":
detectorProps.falseAlarmRate = 0.01 #x-axis
print("falseAlarmRate = {0}".format(detectorProps.falseAlarmRate))
else:
detectorProps.recall = 0.90 #y-axis
print("recall = {0}".format(detectorProps.recall))
results = easy.detect( detector, rocZip, runset, detectorProperties = detectorProps)
easy.printResults( results )
else:
###############################################################
# Without background data
'''
Execute jousting for generating ROC points
'''
contenders = []
for nWord in list_nWord:
c1 = evaluate.Contender("bowROC_oneclass_"+str(nWord))
c1.trainerString = strTrainer
c1.detectorString = strDetector
trainer = easy.getTrainer(c1.trainerString)
trainerProps = easy.getTrainerProperties(trainer)
trainerProps.props["RejectClassStrategy"] = "ignore"
rocZip = easy.makeROCdata(rocData_optimal)
'''
Extract only optimal ROC points
'''
detector = easy.getDetector(strDetector)
detectorProps = easy.getDetectorProperties(detector)
#specify either a desired falseAlarmRate or recall, but not both:
priority = "recall"
if priority == "falseAlarmRate":
detectorProps.falseAlarmRate = 0.01 #x-axis
print("falseAlarmRate = {0}".format(detectorProps.falseAlarmRate))
else:
detectorProps.recall = 0.90 #y-axis
print("recall = {0}".format(detectorProps.recall))
results = easy.detect(detector,
rocZip,
runset,
detectorProperties=detectorProps)
easy.printResults(results)
else:
###############################################################
# Without background data
'''
Execute jousting for generating ROC points
'''
contenders = []
for nWord in list_nWord:
c1 = evaluate.Contender("bowROC_oneclass_" + str(nWord))
c1.trainerString = strTrainer
c1.detectorString = strDetector
trainer = easy.getTrainer(c1.trainerString)
trainerProps = easy.getTrainerProperties(trainer)
def crossValidate( contender, runset, folds=10, printVerbose=False ):
'''Returns summary statistics tp, fp, tn, fn, recall, trueNegRate,
and a detailed matrix of results with one row for
each fold, and one column each for true positive, false
positive, true negative, and false negative counts'''
# sanity checks:
# only positive and negative purposes,
# count number of entries for each purpose
runset_pos = asList( runset, purpose="pos" )
runset_neg = asList( runset, purpose="neg" )
num_items = ( len(runset_pos), len(runset_neg) )
# check that there are no other purposes
all_items = len( asList( runset ) )
if sum(num_items)!=all_items:
raise RuntimeError("crossValidate can only handle Positive and Negative purposes")
if min(num_items)<2:
raise RuntimeError("need more than 1 labeled item per purpose to cross validate")
# make sure there are enough items for xval to make sense
if folds>min(num_items):
print("warning: cannot do "+folds+"-fold validation with only "+str(num_items)+" labeled items")
folds = min(num_items)
# calculate the size of the training and evaluation sets.
# if the number of labeled items in the runset divided
# by the number of folds isn't an even
# division, use more items for the evaluation
chunksize = (int(math.floor( num_items[0]/folds )), int(math.floor( num_items[1]/folds )))
trainsize = (chunksize[0] * (folds-1), chunksize[1] * (folds-1))
evalsize = (num_items[0]-trainsize[0], num_items[1]-trainsize[1])
print( "Will perform a {0}-fold cross-validation with {1} training samples and "
"{2} evaluation samples".format( folds, trainsize, evalsize ) )
# randomize the order of the elements in the runset, once and for all folds
rndidx = ( range( num_items[0] ), range( num_items[1] ) )
random.shuffle( rndidx[0] ) # shuffles items in place
random.shuffle( rndidx[1] ) # shuffles items in place
#confusionTables = numpy.empty( [folds, 5], dtype=int )
confusionTables = []
for fold in range( folds ):
# split the runset
trainset, evalset = splitRunSet( runset_pos, runset_neg, fold, chunksize, evalsize, rndidx )
print( "-------- fold number {0} --------".format(fold) )
# training
print( "---- training:" )
easy.printRunSetInfo( trainset, printArtifacts=printVerbose )
trainer = contender.getTrainer()
model = easy.train( trainer, trainset,
trainerProperties=contender.trainerProps)
# detection
print( "---- evaluation:" )
easy.printRunSetInfo( evalset, printArtifacts=printVerbose )
detector = contender.getDetector()
detections = easy.detect( detector, model, evalset,
detectorProperties=contender.detectorProps )
confusionTables.append( \
getConfusionTable( detections, origSet=evalset, foundMap=contender.foundMap ))
# calculate statistics of our tuple TestResult,nores
sumTestResult = TestResult()
sumNoRes = 0;
for entry in confusionTables:
sumTestResult.tp += entry[0].tp
sumTestResult.tn += entry[0].tn
sumTestResult.fp += entry[0].fp
sumTestResult.fn += entry[0].fn
sumNoRes += entry[1]
r = EvaluationResult(folds, sumTestResult, sumNoRes, detail=None, name=contender.name)
return r
# Make sure all files in the RunSet are available on the remote site;
# it is the client's responsibility to upload them if not.
# The putResult contains information about which files were actually transferred.
#
fileserver = easy.getFileServer( "FileService:default -p 10110 " + host )
putResult = easy.putAllFiles( fileserver, rs1 )
modelfile = "detectors/haarcascade_frontalface_alt.xml"
if not fileserver.exists( easy.getCvacPath(modelfile) ):
easy.putFile( fileserver, easy.getCvacPath(modelfile) )
#
# detect remotely: note the host specification
#
print("------- Remote detection, local result display: -------")
detector = easy.getDetector( "OpenCVCascadeDetector:default -p 10102 "+host )
results = easy.detect( detector, modelfile, rs1 )
easy.printResults( results )
#
# Example 2:
# Train on a remote machine, obtain the model file, and test locally.
# Assume the files are on the remote machine, or transfer with putAllFiles.
#
trainer = easy.getTrainer( "bowTrain:default -p 10103 "+ host) # remote
trainset = easy.createRunSet( "trainImg" );
trainedModel = easy.train( trainer, trainset )
easy.getFile( fileserver, trainedModel.file ) # downloads the model from remote
print("{0}".format(trainedModel))
detector = easy.getDetector( "bowTest:default -p 10104" ) # local service
testset = easy.createRunSet("testImg","UNPURPOSED" )
results = easy.detect( detector, trainedModel, testset )
def test_cvacdatadir(self):
print("testing cvac data dir")
if sys.platform == 'win32':
datadir = os.getenv("CVAC_DATADIR", None)
datadir = datadir.replace("/", "\\")
easy.CVAC_DataDir = datadir
print("Testing using back slashes")
print("Using CVAC_DATADIR as " + datadir)
testset = []
easy.misc.searchDir(testset, datadir + '\\testImg', recursive=True, video=False, image=True)
runset = cvac.RunSet()
easy.addToRunSet(runset, testset, 'pos')
modelfile = 'detectors/bowUSKOCA.zip'
detector = easy.getDetector("BOW_Detector")
props = easy.getDetectorProperties(detector)
props.verbosity = 3
results = easy.detect(detector, modelfile, runset,
detectorProperties=props)
easy.printResults(results)
else:
print("Skipping back slash test on this platform")
#run it again with all forward slashes
datadir = os.getenv("CVAC_DATADIR", None)
datadir = datadir.replace("\\", "/")
print("Testing using all forward slashes")
print("Using CVAC_DATADIR as " + datadir)
easy.CVAC_DataDir = datadir
testset = []
easy.misc.searchDir(testset, datadir + '/testImg', recursive=True, video=False, image=True)
runset = cvac.RunSet()
easy.addToRunSet(runset, testset, 'pos')
modelfile = 'detectors/bowUSKOCA.zip'
detector = easy.getDetector("BOW_Detector")
props = easy.getDetectorProperties(detector)
props.verbosity = 3
results = easy.detect(detector, modelfile, runset,
detectorProperties=props)
easy.printResults(results)
#run it for forward slashes and relative path
origDir = datadir
curDir = os.getcwd()
curDir = curDir.replace("\\", "/")
datadir = datadir.replace("\\", "/")
print("using relative paths for " + curDir + " in data dir " + datadir)
if datadir.startswith(curDir):
datadir = datadir[len(curDir)+1:]
easy.CVAC_DataDir = datadir
print("Using CVAC_DataDir as " + datadir)
testset = []
easy.misc.searchDir(testset, origDir + '/testImg', recursive=True, video=False, image=True)
runset = cvac.RunSet()
easy.addToRunSet(runset, testset, 'pos')
modelfile = origDir + '/detectors/bowUSKOCA.zip'
detector = easy.getDetector("BOW_Detector")
props = easy.getDetectorProperties(detector)
props.verbosity = 3
results = easy.detect(detector, modelfile, runset,
detectorProperties=props)
easy.printResults(results)
else:
RuntimeError("Bad datadir")
# pick a subset: all license plates
license_plates = categories['license plate']
print("There are {0} license plate labels.".format( len(license_plates) ))
# another subset: all labels starting with "car"
cars = []
for key in categories.keys():
if key.startswith("car"):
cars.append( categories[key] )
print("There are {0} car-related labels.".format( len(cars) ))
# Note that Labels are cached in the CorpusServer, but the corpus currently
# needs to re-mirror if the CorpusServer is restarted because Labels are
# not stored to disk. Images are stored to disk.
quit() # done for now
# Train a detector on license plates
trainer = easy.getTrainer( "BOW_Trainer:default -p 10103 ")
trainset = easy.createRunSet( license_plates, "pos" )
easy.printRunSetInfo( trainset )
licenseplateModel = easy.train( trainer, trainset )
# test the license plate detector on the known locations of cars;
# this will only try to detect within the boundaries of cars.
testset = easy.createRunSet( cars, "unpurposed" )
detector = easy.getDetector( "BOW_Detector:default -p 10104" )
results = easy.detect( detector, licenseplateModel, testset )
printResults( results )
import sys, os thisPath = os.path.dirname(os.path.abspath(__file__)) srcPath = os.path.abspath(thisPath+"/../../lib/python") sys.path.append(srcPath) import paths import cvac import easy # # First, a teaser for detection: # detector = easy.getDetector( "bowTest:default -p 10104" ) results = easy.detect( detector, "detectors/bowUSKOCA.zip", "testImg/TestCaFlag.jpg" ) easy.printResults( results ) # # Second, a quick way to train a detector. The resulting model # can be used in place of the detector above. # # TODO: currently breaks because Caltech101 doesn't get extracted as expected #categories, lablist = easy.getDataSet( "corpus/CvacCorpusTest" ) # categories, lablist = easy.getDataSet( "corpus/Caltech101.properties", createMirror=False ) #easy.printCategoryInfo( categories ) #runset = easy.createRunSet( categories["car_side"] ) #trainer = easy.getTrainer( "bowTrain:default -p 10103" ) #carSideModel = easy.train( trainer, runset ) #
#
# Create a training set from one sample each of 9 corporate logos
#
trainset1 = easy.createRunSet("corporate_logos")
# train, round 1
trainer = easy.getTrainer("BOW_Trainer")
model1 = easy.train(trainer, trainset1)
# evaluate the model on a separate test set, images and videos
# in DataDir/testImg
testset1 = easy.createRunSet("testImg", "UNPURPOSED")
easy.printRunSetInfo(testset1)
detector = easy.getDetector("BOW_Detector")
result1 = easy.detect(detector, model1, testset1)
easy.printResults(result1)
# sort the images from the testdata1 folder into subfolders of
# "testresults1" corresponding to the found labels;
# if multiple labels were found per original, consider only
# the label with the highest confidence
easy.sortIntoFolders(result1, outfolder="testresults1", multi="highest")
# Now manually sort through the created testresults1 and move
# _incorrectly_ classified samples into correctly labeled subfolders
# of a new folder "corporate_logos_round2". Found labels on locations
# that are not one of the 9 logos have to be sorted into a 10th class
# (subfolder), generally called the "reject" class.
#
# We simulate this manual process here. Note that the new folder needs to be
# Ex. for image files in a directory
tFileSuffixes = ["jpg","bmp"]
directory_images = cvac.PurposedDirectory(cvac.Purpose(cvac.PurposeType.MULTICLASS,1),
cvac.DirectoryPath("testImg"),
tFileSuffixes,
1)
runset.purposedLists.append(directory_images)
#===============================================================================
# # If you have video files, you may comment out this part for testing the video files
# # Ex. for video files in a directory
# tFileSuffixes = ["mpg","mpeg","avi"]
# directory_videos = cvac.PurposedDirectory(cvac.Purpose(cvac.PurposeType.MULTICLASS,1),
# cvac.DirectoryPath("trainVideo"),
# tFileSuffixes,
# 1)
# runset.purposedLists.append(directory_videos)
#===============================================================================
host = "-h localhost"
detector = easy.getDetector( "RSItrTest_Detector:default -p 10109 " + host )
modelfile = '' #empty detectorData only for testing
results = easy.detect( detector, modelfile, runset )
#results = easy.detect( detector, trainedModel, runset )
#easy.printResults( results, foundMap=classmap )
easy.printResults( results)
quit()
# pick a subset: all license plates
license_plates = categories['license plate']
print("There are {0} license plate labels.".format(len(license_plates)))
# another subset: all labels starting with "car"
cars = []
for key in categories.keys():
if key.startswith("car"):
cars.append(categories[key])
print("There are {0} car-related labels.".format(len(cars)))
# Note that Labels are cached in the CorpusServer, but the corpus currently
# needs to re-mirror if the CorpusServer is restarted because Labels are
# not stored to disk. Images are stored to disk.
quit() # done for now
# Train a detector on license plates
trainer = easy.getTrainer("BOW_Trainer:default -p 10103 ")
trainset = easy.createRunSet(license_plates, "pos")
easy.printRunSetInfo(trainset)
licenseplateModel = easy.train(trainer, trainset)
# test the license plate detector on the known locations of cars;
# this will only try to detect within the boundaries of cars.
testset = easy.createRunSet(cars, "unpurposed")
detector = easy.getDetector("BOW_Detector:default -p 10104")
results = easy.detect(detector, licenseplateModel, testset)
printResults(results)
0, 0)
singleImages.labeledArtifacts.append(cvac.Labelable(0, tlab, tsub))
# Ex. for image files in a directory
tFileSuffixes = ["jpg", "bmp"]
directory_images = cvac.PurposedDirectory(
cvac.Purpose(cvac.PurposeType.MULTICLASS, 1),
cvac.DirectoryPath("testImg"), tFileSuffixes, 1)
runset.purposedLists.append(directory_images)
#===============================================================================
# # If you have video files, you may comment out this part for testing the video files
# # Ex. for video files in a directory
# tFileSuffixes = ["mpg","mpeg","avi"]
# directory_videos = cvac.PurposedDirectory(cvac.Purpose(cvac.PurposeType.MULTICLASS,1),
# cvac.DirectoryPath("trainVideo"),
# tFileSuffixes,
# 1)
# runset.purposedLists.append(directory_videos)
#===============================================================================
host = "-h localhost"
detector = easy.getDetector("RSItrTest_Detector:default -p 10109 " + host)
modelfile = '' #empty detectorData only for testing
results = easy.detect(detector, modelfile, runset)
#results = easy.detect( detector, trainedModel, runset )
#easy.printResults( results, foundMap=classmap )
easy.printResults(results)
quit()
# Make sure all files in the RunSet are available on the remote site;
# it is the client's responsibility to upload them if not.
# The putResult contains information about which files were actually transferred.
#
print("------- Remote detection, local result display: -------")
fileserver = easy.getFileServer("PythonFileService:default -p 10111 " + host)
putResult = easy.putAllFiles(fileserver, rs1)
modelfile = "detectors/haarcascade_frontalface_alt.xml"
if not fileserver.exists(easy.getCvacPath(modelfile)):
easy.putFile(fileserver, easy.getCvacPath(modelfile))
#
# detect remotely: note the host specification
#
detector = easy.getDetector("OpenCVCascadeDetector:default -p 10102 " + host)
results = easy.detect(detector, modelfile, rs1)
easy.printResults(results)
#
# Example 2:
# Train on a remote machine, obtain the model file, and test locally.
# Assume the files are on the remote machine, or transfer with putAllFiles.
# If no local services are installed, this will be skipped.
#
print("------- Remote training, local detection: -------")
try:
detector = easy.getDetector(
"BOW_Detector:default -p 10104") # local service
trainer = easy.getTrainer("BOW_Trainer:default -p 10103 " + host) # remote
trainset = easy.createRunSet("trainImg")
trainedModel = easy.train(trainer, trainset)
''' Easy! mini tutorial Kalman tracker via Matlab Bridge matz 10/2014 ''' import easy # # create a test set as in the Matlab sample at # http://www.mathworks.com/help/vision/ug/object-tracking.html # You need to have the Matlab vision toolbox installed. The # singleball.avi movie is here: # YourMatlabProgramPath/toolbox/vision/visiondemos/singleball.avi # You should move it to the Easy! data directory for the demo # runset = easy.createRunSet( "singleball.avi" ) # Optionally, specify a folder that contains a sequence of the video's # frames as image files # runset = easy.createRunSet( "singleball.avi_frames" ) # # evaluate your tracking algorithm with a common match scoring method # tracker = easy.getDetector( "KalmanTracker" ) results = easy.detect( tracker, None, runset ) easy.printResults( results ) easy.printLabeledTrack( results )
#
# Connect to the trainer for a Bag of Words algorithm, then
# train with the given runset
#
print("starting training, this might take a few minutes...")
trainer = easy.getTrainer("BOW_Trainer")
trainedModel = easy.train(trainer, trainset)
#
# Display information about the file in which the model is stored;
# this is generally no concern to algorithm users and only of
# interest to algorithm developers since it is algorithm-specific
#
zipfname = easy.getFSPath(trainedModel)
print("{0}".format(zipfname))
zipf = zipfile.ZipFile(zipfname)
print("Training model stored in file {0}".format(zipfname))
# print("file contents:\n{0}".format(zipf.namelist()))
#
# test the trained model on a separate set of images
#
print("==== Test runset: ====")
testset = easy.createRunSet("testImg")
easy.printRunSetInfo(testset, printLabels=True)
detector = easy.getDetector("BOW_Detector")
results = easy.detect(detector, trainedModel, testset)
print("==== Results: ====")
easy.printResults(results)
# The tracker needs a labeled location that defines the location
# within the first frame of the video of the item we want to track.
# In this we define a bounding box around the toy soldiers initial location.
vlab = easy.getLabelable( vfile, labelText="soldier" )
labloc = cvac.BBox(255, 195,70,120)
lab = cvac.LabeledLocation()
lab.loc = labloc
lab.confidence = 0
lab.sub = vlab.sub
easy.addToRunSet(runset, lab, easy.getPurpose('pos'))
# Setting the verbosity level gives us some feedback while the tracking is done.
props = easy.getDetectorProperties(detector)
props.verbosity = 7
# A model file is not required for this detector
modelfile = None
results = easy.detect( detector, modelfile, runset, detectorProperties = props)
# <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
'''
Now we run it again this time setting up a callback to display the
tracking frames on the images.
'''
class MyDetectorCallbackReceiverI(easy.DetectorCallbackReceiverI):
def __init__(self):
import easy
easy.DetectorCallbackReceiverI.__init__(self)
def foundNewResults(self, r2, current=None):
easy.DetectorCallbackReceiverI.foundNewResults(self, r2, current)
# Draw the result video frame and the bounding box
correct = easy.isProperRunSet(trainset, deleteInvalid=True)
if not correct:
print("failed Integrity test!!!!")
exit()
#trainset = easy.createRunSet( categories, purpose=posPurpose );
#easy.printRunSetInfo( trainset, printLabels=True )
easy.printRunSetInfo( trainset )
#
# Connect to the trainer for a Bag of Words algorithm, then
# train with the given runset
#
#trainer = easy.getTrainer( "BOW_Trainer")
trainer = easy.getTrainer( "BOW_Trainer:default -p 10103")
trainedModel = easy.train( trainer, trainset )
zipfname = easy.getFSPath( trainedModel )
print("Training model stored in file {0}".format( zipfname ))
#
# test the trained model on a separate set of images
#
print("==== Test runset: ====")
testset = easy.createRunSet( "testImg" )
easy.printRunSetInfo( testset, printLabels=True )
detector = easy.getDetector( "BOW_Detector" )
results = easy.detect( detector, trainedModel, testset )
print("==== Results: ====")
easy.printResults( results )
# # Create a training set from one sample each of 9 corporate logos # trainset1 = easy.createRunSet( "corporate_logos" ) # train, round 1 trainer = easy.getTrainer( "BOW_Trainer") model1 = easy.train( trainer, trainset1 ) # evaluate the model on a separate test set, images and videos # in DataDir/testdata1 testset1 = easy.createRunSet( "testImg", "UNPURPOSED" ) easy.printRunSetInfo( testset1 ) detector = easy.getDetector( "BOW_Detector" ) result1 = easy.detect( detector, model1, testset1 ) easy.printResults(result1) # sort the images from the testdata1 folder into subfolders of # "testresults1" corresponding to the found labels; # if multiple labels were found per original, consider only # the label with the highest confidence easy.sortIntoFolders( result1, outfolder="testresults1", multi="highest") # Now manually sort through the created testresults1 and move # _incorrectly_ classified samples into correctly labeled subfolders # of a new folder "corporate_logos_round2". Found labels on locations # that are not one of the 9 logos have to be sorted into a 10th class # (subfolder), generally called the "reject" class. # # We simulate this manual process here. Note that the new folder needs to be
# labels. To use them for evaluation, let's assign the same purpose
# to syntactically different but semantically identical labels.
# Because we don't specify it, this guesses the specific Purpose that
# is assigned to the labels.
# Also obtain this mapping from Purpose to label name, called "classmap."
rs1 = easy.createRunSet( categories1['CA_flag']+categories2['ca'], "0" )
easy.addToRunSet( rs1, categories1['KO_flag']+categories2['kr'], "1" )
easy.addToRunSet( rs1, categories1['US_flag']+categories2['us'], "2" )
print("\n=== The Corpora combined into one RunSet: ===");
easy.printRunSetInfo( rs1 )
# A runset can be used for training or for testing
print("------- Bag of Words results for corporate logos: -------")
detector = easy.getDetector( "BOW_Detector" )
modelfile = "detectors/bowUSKOCA.zip"
results1 = easy.detect( detector, modelfile, rs1 )
print("Note that both original and found labels are printed:")
easy.printResults( results1 )
# Print again, this time replacing the found labels with a double
# mapping from foundLabel -> guessed Purpose -> classmap label;
# Note that this fails if multiple original labels mapped to the same
# Purpose.
wait()
print("------- Same results, but found labels replaced with guessed original labels: -------")
easy.printResults( results1, foundMap=rs1['classmap'], inverseMap=True )
# Print again, this time replacing all labels with their assigned
# purposes, bot original and found labels. Note the "identical label"
# matches.
wait()
'''
import easy
# obtain a reference to a Bag of Words (BOW) detector
detector = easy.getDetector( "BOW_Detector" )
# a model for distinguishing Canadian, Korean, and US flags,
# trained previously with a BOW-specific trainer and stored in a file
modelfile = "detectors/bowUSKOCA.zip"
# a test image; the location is relative to the "CVAC.DataDir"
imgfile = "testImg/TestKrFlag.jpg"
# apply the detector type, using the model and testing the imgfile
results = easy.detect( detector, modelfile, imgfile )
# you can print the results with Easy!'s pretty-printer;
# we will explain the meaning of the "unlabeled" and the number
# of the found label later.
# or you can always print variables directly (uncomment the next line):
# print("{0}".format( results ))
print("------- Bag of Words results for flags: -------")
easy.printResults( results )
# let's try a different model, pre-trained not for
# flags but for various corporate logos
print("------- Bag of Words results for corporate logos: -------")
modelfile = "detectors/bowCorporateLogoModel.zip"
imgfile = "corporate_logos/shell/shell2.png"
results = easy.detect( detector, modelfile, imgfile )
'''
Easy! mini tutorial
Kalman tracker via Matlab Bridge
matz 10/2014
'''
import easy
#
# create a test set as in the Matlab sample at
# http://www.mathworks.com/help/vision/ug/object-tracking.html
# You need to have the Matlab vision toolbox installed. The
# singleball.avi movie is here:
# YourMatlabProgramPath/toolbox/vision/visiondemos/singleball.avi
# You should move it to the Easy! data directory for the demo
#
runset = easy.createRunSet("singleball.avi")
# Optionally, specify a folder that contains a sequence of the video's
# frames as image files
# runset = easy.createRunSet( "singleball.avi_frames" )
#
# evaluate your tracking algorithm with a common match scoring method
#
tracker = easy.getDetector("KalmanTracker")
results = easy.detect(tracker, None, runset)
easy.printResults(results)
easy.printLabeledTrack(results)
# labels. To use them for evaluation, let's assign the same purpose
# to syntactically different but semantically identical labels.
# Because we don't specify it, this guesses the specific Purpose that
# is assigned to the labels.
# Also obtain this mapping from Purpose to label name, called "classmap."
rs1 = easy.createRunSet(categories1['CA_flag'] + categories2['ca'], "0")
easy.addToRunSet(rs1, categories1['KO_flag'] + categories2['kr'], "1")
easy.addToRunSet(rs1, categories1['US_flag'] + categories2['us'], "2")
print("\n=== The Corpora combined into one RunSet: ===")
easy.printRunSetInfo(rs1)
# A runset can be used for training or for testing
print("------- Bag of Words results for corporate logos: -------")
detector = easy.getDetector("BOW_Detector")
modelfile = "detectors/bowUSKOCA.zip"
results1 = easy.detect(detector, modelfile, rs1)
print("Note that both original and found labels are printed:")
easy.printResults(results1)
# Print again, this time replacing the found labels with a double
# mapping from foundLabel -> guessed Purpose -> classmap label;
# Note that this fails if multiple original labels mapped to the same
# Purpose.
wait()
print(
"------- Same results, but found labels replaced with guessed original labels: -------"
)
easy.printResults(results1, foundMap=rs1['classmap'], inverseMap=True)
# Print again, this time replacing all labels with their assigned
# purposes, bot original and found labels. Note the "identical label"