def __init__(self):
self.CONF = 0.7
self.classifier = classify.Classify()
self.preprocessor = preprocess.PreProcessor()
self.class_names = self.classifier.get_class_names()
self.cache = cache.Cache(max_size=10)
self.faceCascade = cv2.CascadeClassifier(
'haarcascade_frontalface_default.xml')
def process(df, frac_str, type_str, dimensions, balance=True):
"""
Apply data transformations/dimensionality reduction
:param frac_str: string denoting percentage of dataset operated on
:param dimensions: List of int values to reduce dimensions to
:param balance: Boolean apply class re-sampling to resolve imbalance
:type df: Pandas dataframe
:type type_str: training or test data label
"""
preprocessor = preprocess.PreProcessor()
data_arr, labels = do_transforms(df, type_str, frac_str, preprocessor)
for dimension in dimensions:
print("Applying dimensionality reduction...")
data_lsa = preprocessor.truncate(data_arr, dimension)
print("Reduction produced a: {}".format(type(data_lsa)))
print("With shape: {}".format(data_lsa.shape))
print("Saving reduced data...")
joblib.dump(data_lsa,
'data/{}/{}-data-dm-{}.gz'.format(frac_str, type_str,
dimension),
compress=3)
del data_lsa
del data_arr
if balance:
for dimension in dimensions:
print("Applying class re-sampling to array with dimensions: {}...".
format(dimension))
data_arr = joblib.load('data/{}/{}-data-dm-{}.gz'.format(
frac_str, type_str, dimension))
data_rs, label_rs = preprocessor.balance(data_arr, labels)
print("Re-sampling produced a: {}".format(type(data_rs)))
print("With shape: {}".format(data_rs.shape))
print("Label shape: {}".format(label_rs.shape))
print("Saving prepared data...")
joblib.dump(data_rs,
'data/{}/{}-data-rs-{}.gz'.format(
frac_str, type_str, dimension),
compress=3)
joblib.dump(label_rs,
'data/{}/{}-label-rs-{}.gz'.format(
frac_str, type_str, dimension),
compress=3)
del data_rs, label_rs
import classify
import sys
import cv2
import preprocess
import time
classifier = classify.Classify()
preprocessor = preprocess.PreProcessor()
camera = cv2.VideoCapture(0)
i = 0;
start = time.time()
prediction_out_dir = "prediction_output_images/"
print("start_time ", start)
while True:
return_value, image = camera.read()
# print("time new capture:", time.asctime( time.localtime(time.time()) ))
# cv2.imwrite('opencv'+str(i)+'.png', image) commented by anuj
cv2.imwrite(prediction_out_dir + 'opencv.png', image)
# print("time image written:", time.asctime( time.localtime(time.time()) ))
# bb = (preprocessor.align('opencv'+str(i)+'.png')) commented by anuj
bb = (preprocessor.align(prediction_out_dir + 'opencv.png'))
# print("time alignment done: ", time.asctime( time.localtime(time.time()) ))
if bb.any() == False:
pass
else:
cv2.rectangle(image, (bb[0],bb[1]), (bb[2],bb[3]), (0, 255, 0), 5)