def __init__(self, result_dir, sampleModel: SamplesModel):
self.workDir = propsmain.get('workdirpath')
self.resultDir = os.path.join(self.workDir, result_dir)
self.sampleCount = sampleModel.positives
self.xAngle = sampleModel.x_angle
self.yAngle = sampleModel.y_angle
self.zAngle = sampleModel.z_angle
self.iDev = sampleModel.max_dev
self.w = sampleModel.w
self.h = sampleModel.h
self.perlScript = os.path.join(
BASE_DIR, 'earTrainer/scripts/createtrainsamples.pl')
self.samplesRootPath = propsmain.get('samplespath')
self.samplesDir = sampleModel.samples_dir
self.negativesDir = 'negatives'
self.imageFormat = propsmain.get('imageformat')
self.samplesModel = sampleModel
self.positivesDat = 'None'
self.negativesDat = 'None'
if not os.path.exists(self.workDir):
os.mkdir(self.workDir)
if os.path.exists(self.resultDir):
shutil.rmtree(self.resultDir)
os.mkdir(self.resultDir)
def __init__(self, trainModel: TrainerModel):
self.resultDir = trainModel.name
#ukladanie do vedlajsieho precinka
self.resultDir = propsmain.get('resultsdir')
self.workDir = propsmain.get('workdirpath')
self.samplesDir = trainModel.positives.name
self.numPos = trainModel.positives.positives
self.numNeg = trainModel.negatives
self.numStages = trainModel.num_stages
self.precalcValBuf = trainModel.precalcValBuf
self.precalcIdxBuf = trainModel.precalcIdxBuf
self.numThreads = trainModel.numThreads
self.acceptBreak = trainModel.acceptanceBreak
self.w = trainModel.positives.w
self.h = trainModel.positives.h
self.bt = trainModel.bt
self.minHitRate = trainModel.minHitRate
self.maxFalseAlarm = trainModel.maxFalseAlarm
self.weightTrimRate = trainModel.weightTrimRate
self.maxDepth = trainModel.maxDepth
self.maxWeakCount = trainModel.maxWeakCount
self.featureType = trainModel.featureType
self.trainerModel = trainModel
self.mode = trainModel.mode
self.opencv_trainer = 'opencv_traincascade.exe'
if platform == 'linux':
self.opencv_trainer = self.opencv_trainer[:-4]
def get_all_xmls():
# all_xmls = glob.glob(BASE_DIR + '/earDetector/xml/*.xml')
# xmls = list()
# for one_xml in all_xmls:
# xmls.append((os.path.basename(one_xml), one_xml))
xmls = get_files(os.path.join(properties.get('testerdir'), 'xmls/'),
'.xml')
return xmls
def __init__(self,samplesDir='samples/',descriptor='descriptor.txt'):
self.root = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
self.root = propsmain.get('testerdir')
self.descriptor_dir = os.path.join(self.root,'descriptors/')
if not os.path.exists(self.descriptor_dir):
os.makedirs(self.descriptor_dir)
self.results = os.path.join(self.root,'results/')
self.samples = os.path.join(self.root, samplesDir)
if not os.path.exists(os.path.join(self.descriptor_dir,descriptor)):
raise FileNotFoundError('Descriptor '+self.descriptor+' not found in '+self.descriptor_dir)
self.descriptor = os.path.join(self.descriptor_dir,descriptor)
def __init__(self,
xml_ear_file='',
descriptor_name='descriptor.txt',
trainer_name='default',
custom=False,
samples_dir='samples/'):
self.root_dir = os.path.dirname(
os.path.dirname(os.path.abspath(__file__)))
self.custom = custom
self.root_dir = propsmain.get('testerdir')
self.xmls_dir = os.path.join(self.root_dir, 'xmls/')
if not os.path.exists(self.xmls_dir):
os.makedirs(self.xmls_dir)
self.detector = None
self.xml_file = ''
self.xml_ear_file = xml_ear_file
if xml_ear_file is not '' and not custom:
#copy to xmls dir
new_name = trainer_name + '.xml'
copyfile(xml_ear_file, os.path.join(self.xmls_dir, new_name))
self.xml_file = os.path.join(self.xmls_dir, new_name)
#os.rename(os.path.join(self.xmls_dir,'cascade.xml'),self.xml_file)
self.detector = cv2.CascadeClassifier(self.xml_file)
self.result_dir = os.path.join(self.root_dir, 'results/')
else:
self.detector = cv2.CascadeClassifier(self.xml_ear_file)
self.result_file_name = 'results.txt'
self.result_ear_not_found = 'not_found.txt'
self.testing_samples = os.path.join(self.root_dir, samples_dir)
self.descriptor = descriptor_name
self.penalty = 0
self.comparator = Comparator(samplesDir=self.testing_samples,
descriptor=self.descriptor)
def __init__(self, xmlfile, type):
self.detector = cv2.CascadeClassifier(
os.path.join(properties.get('testerdir'), 'xmls', xmlfile))
self.equalizer = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(4, 4))
self.result_images = list()
self.type = type
def get_all_descriptors():
return get_files(os.path.join(properties.get('testerdir'), 'descriptors/'),
'.txt')
def get_all_test_dir():
return get_dirs(properties.get('testerdir'),
['xmls', 'results', 'descriptors'])
def get_all_samples_dir():
return get_dirs(properties.get('samplespath'))