def the_method():
if request.headers['Content-Type'] == 'text/plain':
return "Text Message: " + request.data
elif request.headers['Content-Type'] == 'application/json':
return "JSON Message: " + json.dumps(request.json)
elif request.headers['Content-Type'] == 'application/octet-stream':
print(request.data)
f = open('./binary', 'wb')
f.write(request.data)
f.close()
return "Binary message written!"
else:
f = request.files['binarydata']
f.seek(0)
my_bytes = f.read()
index = 0
matrix = []
row = []
for byte in my_bytes:
normalized = (float(byte) - 0) / (255 - 0)
row.append(normalized)
if (index % row_size == (row_size - 1)):
matrix.append(row)
row = []
index = index + 1
np_array = np.array([matrix], dtype=np.float64)
print(np_array[0].shape)
#print("np_array[0]. Type : {} | Value : {}".format(type(np_array[0][0][0]),np_array[0][0][0]))
neural_network = NeuralNet(row_size, row_size)
neural_network.load_model('StoredModel.modelconfig')
arg_max, max_val = neural_network.predict_element(np_array)
lb = labels_map[arg_max]
print('ArgMax : {} | MaxVal : {} |Label : {}'.format(
arg_max, max_val, lb))
result = {
'Confidence': max_val.item(),
'Predicted': arg_max.item(),
'Label': lb
}
return json.dumps(result)
import numpy as np
import sys
from neural import NeuralNet
from dataset import get_all_categories_shuffled, numpy_array_from_file, prepare_data
labels_dictionary = {'Circle': 0, 'L': 1, 'RightArrow': 2}
labels_map = ['Circle', 'L', 'RightArrow']
#Replace with your root folder that holds the dataset of images
all_bytes, all_labels = prepare_data("Datasets/DatasetSample0/Images",
labels_dictionary, 56, 56)
nn = NeuralNet(56, 56)
nn.build_layers()
nn.fit_data(training_data=all_bytes,
training_labels=all_labels,
epochs=10,
accuracy=0.999)
nn.save_model('easter_egg_ahlabikyafraise_')
#nn.load_model('easter_egg_ahlabikyafraise_')
image_path = sys.argv[1]
test_image = numpy_array_from_file(image_path)
test_image = test_image / 255
arg_max, prediction_level = nn.predict_element(test_image)
print('arg_max : {} which is {} with prediction : {}'.format(
arg_max, labels_map[arg_max], prediction_level))