def test_jpeg_rotation(self):
# Make sure all Portrait test images are auto-rotated correctly
for i in range(9):
img_jpg = load_image_file(f"Portrait_{i}.jpg")
ref_img = np.load(f"Portrait_{i}.jpg.npy")
self.assertTrue(np.array_equal(ref_img, img_jpg))
# Make sure all Landscape test images are auto-rotated correctly
for i in range(9):
img_jpg = load_image_file(f"Landscape_{i}.jpg")
ref_img = np.load(f"Landscape_{i}.jpg.npy")
self.assertTrue(np.array_equal(ref_img, img_jpg))
def analyze_color(image_path, candidates=3):
# Reading Image file
img = image_to_numpy.load_image_file(image_path)
img = cv2.resize(img, (180, 180))
ar = np.asarray(img)
# Collapse image to rgb array
ar = np.reshape(ar, newshape=(-1, ar.shape[-1])).astype(float)
# print('finding clusters')
codes, dist = kmeans(ar, NUM_CLUSTERS)
# print('cluster centres:\n', codes)
vecs, dist = vq(ar, codes) # assign codes
counts, bins = np.histogram(vecs, len(codes)) # count occurrences
sorted_idx = np.argsort(counts)[::-1]
res = []
for i in range(candidates):
peak = codes[sorted_idx[i]]
res += [ntc.name(peak)]
return res
def kakao_ocr(image_path: str, api_key: str):
"""
OCR api request example
:param image_path: 이미지파일 경로
:param api_key: 카카오 앱 REST API 키
"""
resize_impath = kakao_ocr_resize(image_path)
if resize_impath is not None:
image_path = resize_impath
headers = {'Authorization': 'KakaoAK {}'.format(api_key)}
image = image_to_numpy.load_image_file(image_path)
image = cv2.imencode('.png', image)[1]
data = image.tobytes()
api_response = requests.post(API_URL,
headers=headers,
files={
"image": data
}).json()
print(api_response)
if resize_impath is not None:
os.remove(resize_impath)
if len(api_response['result']) == 0:
return -1
result = [
token['recognition_words'][0] for token in api_response['result']
]
return ' '.join(result)
def is_valid(self):
if self.__validated is not None:
return self.__validated
try:
if self.url:
req = request.urlopen(self.url)
self.image = image_to_numpy.load_image_file(BytesIO(req.read()))
elif self.image:
base64_data = re.sub('^data:image/.+;base64,', '', self.image)
byte_data = base64.b64decode(base64_data)
image_data = BytesIO(byte_data)
self.image = image_to_numpy.load_image_file(image_data)
self.__validated = True
except Exception as e:
self.__validated = False
return self.__validated
def load_image(image_path):
# img = tf.io.read_file(image_path)
# img = tf.image.decode_jpeg(img, channels=3)
img = image_to_numpy.load_image_file(image_path)
# plt.imshow(img)
# plt.show()
img = tf.image.resize(img, (380, 380)) # (380, 380) for efficient-netB4
img = tf.keras.applications.efficientnet.preprocess_input(img)
return img, image_path
def test_png(self):
# Can we load a non-jpeg file with no metadata?
img_png = load_image_file("Portrait_8.png")
self.assertEqual(img_png.shape, (1800, 1200, 3))
def test_jpeg_no_exif(self):
# Can we load a jpeg with no metadata without crashing?
img_jpg = load_image_file("Portrait_no_exif.jpg")
self.assertEqual(img_jpg.shape, (1200, 1800, 3))
COLOR = [0, 0, 255]
DETECTION_MODEL = "hog"
log = Logger(os.path.basename(__file__))
source_faces = []
start = time.perf_counter()
"""LOADING SOURCE IMAGES"""
log.info(f"loading source images")
for index, filename in enumerate(os.listdir(SOURCE_DATA_DIR)):
if filename.endswith("JPG"):
log.info(f"processing source image {filename}")
img_path = SOURCE_DATA_DIR + filename
# using image_to_numpy to load img file -> fixes image orientation -> face encoding is found
img = image_to_numpy.load_image_file(img_path)
try:
source_img_encoding = face_recognition.face_encodings(img)[0]
log.success("face encoding found")
source_faces.append(source_img_encoding)
except IndexError:
log.error("no face detected")
if (len(os.listdir(SOURCE_DATA_DIR)) -
1) - len(source_faces) != 0: # -1 for .gitignore
log.warn(
f"{str(len(source_faces))} faces found in {str(len(os.listdir(SOURCE_DATA_DIR)))} images"
)
"""MAIN PROCESS"""
log.info(f"Processing dataset")
for index, filename in enumerate(os.listdir(IMAGE_SET_DIR)):
def main():
#----------Load Known values----------#
known_encodings = [] #arrays for encodings and associated names
known_names = []
for filename in os.listdir("./images"): #loop through images folder for known names and images
if filename.startswith("."): #ignore .ds_store and other hiddens
continue
image_file = fr.load_image_file("./images/" + filename)
image_encodings = fr.face_encodings(image_file)[0]
known_encodings.append(image_encodings)
name = filename.split(".")[0] #better hope there are no periods in name
known_names.append(name)
face_locations, face_encodings, face_names = [], [], [] #arrays for processing sent frames
process_frame = 0 #counter to allow processing every n frames
n = 1
#----------Process Captured Frames----------#
for fname in os.listdir("./captured_frames"): #loop through frames stored in captured_frames folder
if fname.startswith("."): #ignore .ds_store and other hiddens
continue
frame = im_np.load_image_file("./captured_frames/" + fname)
RESIZE = 2
#resize for faster processing
small_frame = cv2.resize(frame, (0,0), fx=1/RESIZE, fy=1/RESIZE)
rgb_small = small_frame[:, :, ::-1]
#process every few frames
if process_frame % n == 0:
face_locations = fr.face_locations(rgb_small)
face_encodings = fr.face_encodings(rgb_small, face_locations)
#check each set of encodings for a match
for face_encoding in face_encodings:
matches = fr.compare_faces(known_encodings, face_encoding)
name = "Unknown"
#get "closest" match
face_distances = fr.face_distance(known_encodings, face_encoding)
best_match_index = np.argmin(face_distances)
if matches[best_match_index]:
name = known_names[best_match_index]
#append name to list
if not (name in face_names):
face_names.append(name)
process_frame += 1
tot_names = len(face_names)
if tot_names > 0: #append names to response string
response = "I have found "
c = 1
for name in face_names:
response += name
if c < tot_names:
response += " and "
c += 1
print(response)
else:
print("I have found no one")