def test_example(cv_image, message, jointModelObject, print_message):
# convert cv image into processing format
# TODO: this needs to be corrected
# we do not read a file
# we convert from one format to the other
#image = utils.imageRead(imageFile)
print("Test an Example...")
if cv_image == None:
print_message("Unable to capture an Image.")
return
image = cv_image
# extract color and shape of image
# image_copy = copy.copy(image)
cnt = utils.objectIdentification(cv_image)
[x, y, w, h] = utils.boundingRectangle(cnt)
pixNp = dc.findAllPixels(image, cnt, x, y, w, h)
pixNp = dc.findUniquePixels(pixNp)
# store image data as dictionary
imageData = {'color': pixNp, 'shape': cnt}
print("Printing the size of RGB value " + str(len(pixNp)))
# extract keywords from message
languageObject = lm(message)
[positiveLanguageData, negativeLanguageData] = languageObject.process_content()
print(ctime())
# Now perform the test.
[totalScore, wordRanks, wordScoreDictionary] = jointModelObject.associate_words_example(positiveLanguageData, negativeLanguageData, imageData)
print_message("Score: " + str(totalScore))
print_message(str(wordRanks))
print_message(str(wordScoreDictionary))
print(ctime())
def processImage(self, cv_image):
image = cv_image
# extract color and shape of image
# image_copy = copy.copy(image)
cnt = utils.objectIdentification(cv_image)
[x, y, w, h] = utils.boundingRectangle(cnt)
pixNp = dc.findAllPixels(image, cnt, x, y, w, h)
pixNp = dc.findUniquePixels(pixNp)
# store image data as dictionary
imageData = {'color': pixNp, 'shape': cnt}
print("Total RGB values : " + str(len(pixNp)))
# Return the image information.
return imageData
def add_example(cv_image, message, jointModelObject, print_message, example_count):
# convert cv image into processing format
# TODO: this needs to be corrected
# we do not read a file
# we convert from one format to the other
# image = utils.imageRead(imageFile)
print("Adding an Example...")
if cv_image == None:
print_message("Unable to capture an Image.")
return
image = cv_image
# extract color and shape of image
# image_copy = copy.copy(image)
cnt = utils.objectIdentification(cv_image)
[x, y, w, h] = utils.boundingRectangle(cnt)
pixNp = dc.findAllPixels(image, cnt, x, y, w, h)
pixNp = dc.findUniquePixels(pixNp)
# store image data as dictionary
imageData = {'color': pixNp, 'shape': cnt}
print("Printing the size of RGB value " + str(len(pixNp)))
# extract keywords from message
languageObject = lm(message)
[positiveLanguageData, negativeLanguageData] = languageObject.process_content()
# for each keyword
# add keyword, image pair to joint model
for keyword in positiveLanguageData:
jointModelObject.add_word_example_pair(keyword, imageData, "+")
for keyword in negativeLanguageData:
jointModelObject.add_word_example_pair(keyword, imageData, "-")
# Send ACK to output Node that the concept has been added.
# Pickle the data to store training information.
# Store size is the number of examples after which the model is stored to a pickle file.
store_size = 20
if ((example_count % store_size) == 0):
with open('data/pickle/passive_jointModelObject.pickle', 'wb') as handle:
pickle.dump(jointModelObject, handle)
print_message("Example Object - Concept Added. Number of Examples Added: " + str(example_count))