def infer(model, directory):
""" infer a set of images for classification"""
click.clear()
click.echo()
click.echo(lr.APP_TITLE)
fm = FileManager()
directory = fm.adjustDirectoryPath(directory)
logTrainer = LogisticImageTrainer()
if os.path.exists(lr.LOCAL_SAVED_MODEL_LR):
logTrainer.loadModel(lr.LOCAL_SAVED_MODEL_LR)
print("== model", lr.LOCAL_SAVED_MODEL_LR, "loaded")
it = ImageTools()
imageList = fm.folderFileListOfImages(directory)
if len(imageList) > 0:
for filename in imageList:
code,descr = logTrainer.infer(directory,filename)
out = "== FILE {0:s} Class {1:s}->{2:s}".format(filename, code ,descr)
click.echo(out)
for filename in imageList:
code,descr = logTrainer.infer(directory,filename)
out = "== {0:s} -> {1:s}".format( code ,descr)
it.showImage(directory+filename,out)
else:
click.echo(':: No files in \'%s\' directory' % directory)
else:
print("== model", lr.LOCAL_SAVED_MODEL_LR, "required for predictions do not exist!")
def infer(self, path, filename):
"""Allows to infer on an image, based on a saved model
"""
fm = FileManager()
path = fm.adjustDirectoryPath(path)
it = ImageTools()
newfile = it.resize(path, filename, lr.IMAGE_RESIZE_ROWS,
lr.IMAGE_RESIZE_COLS)
imgArray = []
imgArray.append(it.getImageArray(path, newfile))
os.remove(path + newfile)
prediction = str(self.__model.predict(imgArray)[0])
description = self.__CLassDic[prediction]
return prediction, description
def __loadSet(self, path):
"""Given a directory with images, load all images as set.
Set can be use later as a train or test set
"""
self.__deleteSets()
self.__tempoDataSet = []
self.__tempoLabelSet = []
fm = FileManager()
path = fm.adjustDirectoryPath(path)
imageList = fm.folderFileListOfImages(path)
it = ImageTools()
for filename in imageList:
imgArray = it.getImageArray(path, filename)
parts = filename.split('_')
IdClass = int(parts[0])
self.__tempoDataSet.append(imgArray)
self.__tempoLabelSet.append(IdClass)
