def main(training_set_ratio):
arrows = pd.DataFrame(np.zeros((3, 4), dtype=np.int32),
index=('hollow', 'full', 'thin'),
columns=('down', 'left', 'right', 'up'))
images = common.get_files(common.SAMPLES_DIR)
if images:
for _, filename in images:
arrow_direction, arrow_type = common.arrow_labels(filename)
arrows[arrow_direction][arrow_type] += 1
num_samples = int(arrows.min().min() * training_set_ratio)
print("Samples per type: {}".format(num_samples * 4))
for t, _ in arrows.iterrows():
print("\nProcessing {} arrows...".format(t))
for d in arrows:
candidates = [(p, f) for p, f in images
if common.arrow_labels(f) == (d, t)]
print("{}: {}".format(d, len(candidates)))
training = random.sample(candidates, num_samples)
for path, filename in training:
dst_dir = common.TRAINING_DIR + d + "/"
if not os.path.exists(dst_dir):
os.makedirs(dst_dir)
os.rename(path, dst_dir + filename)
candidates = [c for c in candidates if c not in training]
validation = random.sample(
candidates, int(len(candidates) * VALIDATION_SET_RATIO))
for path, filename in validation:
dst_dir = common.VALIDATION_DIR + d + "/"
if not os.path.exists(dst_dir):
os.makedirs(dst_dir)
os.rename(path, dst_dir + filename)
testing = [c for c in candidates if c not in validation]
for path, filename in testing:
dst_dir = common.TESTING_DIR + d + "/"
if not os.path.exists(dst_dir):
os.makedirs(dst_dir)
os.rename(path, dst_dir + filename)
show_summary()
print("\nFinished!")
def main(training_set_ratio):
common.create_directories()
arrows = pd.DataFrame(np.zeros((3, 4), dtype=np.int32),
index=('round', 'wide', 'narrow'),
columns=('down', 'left', 'right', 'up'))
images = [(p, f) for p, f in common.get_files(common.SAMPLES_DIR)
if f[-5] != 'F']
if images:
for _, filename in images:
arrow_direction, arrow_type = common.arrow_labels(filename)
arrows[arrow_direction][arrow_type] += 1
num_samples = int(arrows.min().min() * training_set_ratio)
print("Samples per type: {}".format(num_samples * 4))
for t, _ in arrows.iterrows():
print("\nProcessing {} arrows...".format(t))
for direction in arrows:
candidates = [(p, f) for p, f in images
if common.arrow_labels(f) == (direction, t)]
print("{}: {}".format(direction, len(candidates)))
training = random.sample(candidates, num_samples)
for path, filename in training:
dst_dir = common.TRAINING_DIR + direction + '/'
os.rename(path, dst_dir + filename)
os.rename(flipped(path), dst_dir + flipped(filename))
candidates = [c for c in candidates if c not in training]
validation = random.sample(
candidates, int(len(candidates) * VALIDATION_SET_RATIO))
for path, filename in validation:
dst_dir = common.VALIDATION_DIR + direction + '/'
os.rename(path, dst_dir + filename)
os.rename(flipped(path), dst_dir + flipped(filename))
testing = [c for c in candidates if c not in validation]
for path, filename in testing:
dst_dir = common.TESTING_DIR + direction + '/'
os.rename(path, dst_dir + filename)
os.rename(flipped(path), dst_dir + flipped(filename))
show_summary()
print("\nFinished!")
def get_summary_matrix(directory):
matrix = pd.DataFrame(np.zeros((4, 5), dtype=np.int32),
index=('round', 'wide', 'narrow', 'total'),
columns=('down', 'left', 'right', 'up', 'total'))
images = common.get_files(directory)
for _, filename in images:
arrow_direction, arrow_type = common.arrow_labels(filename)
matrix[arrow_direction][arrow_type] += 1
matrix['total'][arrow_type] += 1
matrix[arrow_direction]['total'] += 1
matrix['total']['total'] += 1
return matrix
def show_summary():
matrix = pd.DataFrame(np.zeros((4, 5), dtype=np.int32), index=(
'hollow', 'full', 'thin', 'total'), columns=('down', 'left', 'right', 'up', 'total'))
images = common.get_files(common.SAMPLES_DIR)
for _, filename in images:
arrow_direction, arrow_type = common.arrow_labels(filename)
matrix[arrow_direction][arrow_type] += 1
matrix['total'][arrow_type] += 1
matrix[arrow_direction]['total'] += 1
matrix['total']['total'] += 1
print(cf.salmon("Samples summary"))
print(matrix, "\n")
def main(src_subdir, verbose, model_name):
common.create_directories()
src_dir = common.DATA_DIR + src_subdir + "/"
model = tensorflow.keras.models.load_model(common.MODEL_DIR + model_name)
# real (index) x predicted (column)
confusion_matrix = pd.DataFrame(np.zeros((4, 4), dtype=np.int32),
index=('down', 'left', 'right', 'up'),
columns=('down', 'left', 'right', 'up'))
classification_matrix = pd.DataFrame(np.zeros((4, 3)),
index=('down', 'left', 'right', 'up'),
columns=('precision', 'recall', 'f1'))
type_matrix = pd.DataFrame(np.zeros((4, 2), dtype=np.int32),
index=('round', 'wide', 'narrow', 'total'),
columns=('correct', 'incorrect'))
images = common.get_files(src_dir)
print("Processing {} file(s) in {}/...\n".format(len(images), src_subdir))
for path, filename in images:
img = cv2.imread(path, cv2.IMREAD_GRAYSCALE)
data = np.reshape(img, (1, ) + common.INPUT_SHAPE)
prediction = model.predict(data)
class_index = np.argmax(prediction)
predicted_class = common.CLASSES[class_index]
real_class, arrow_type = common.arrow_labels(filename)
if verbose and real_class != predicted_class:
print(path)
print("Expected {} but got {}: {}\n".format(
cf.lightGreen(real_class), cf.lightCoral(predicted_class),
str(prediction[0])))
confusion_matrix[predicted_class][real_class] += 1
if real_class == predicted_class:
type_matrix['correct'][arrow_type] += 1
type_matrix['correct']['total'] += 1
else:
type_matrix['incorrect'][arrow_type] += 1
type_matrix['incorrect']['total'] += 1
print("\n" + cf.sandyBrown("Confusion matrix"))
print(confusion_matrix)
classification_matrix['precision'] = confusion_matrix.apply(precision)
classification_matrix['recall'] = confusion_matrix.apply(recall, axis=1)
classification_matrix['f1'] = classification_matrix.apply(f1, axis=1)
print("\n" + cf.skyBlue("Classification summary"))
print(classification_matrix)
type_matrix['accuracy'] = type_matrix.apply(type_accuracy, axis=1)
print("\n" + cf.plum("Accuracy by type"))
print(type_matrix)
print("\nFinished!")