def run_tf_object_detection_video(input_graph,
image_tensor,
tensor_dict,
path_to_video=None):
if path_to_video == None:
cap = cv2.VideoCapture(0)
else:
cap = cv2.VideoCapture(path_to_video)
if cap.isOpened() is False:
print("Error opening video stream or file")
with input_graph.as_default():
with tf.Session() as sess:
while cap.isOpened():
_, image = cap.read()
image_np = (cv2.cvtColor(image, cv2.COLOR_BGR2RGB)).astype(
np.uint8) #Convert to RGB and convert to uint8
output_dict = run_inference_for_single_image_through_ssd(
sess, image_np, image_tensor, tensor_dict)
dets = post_process_ssd_predictions(image_np,
output_dict,
threshold=0.25)
for cur_det in dets:
boxes = cur_det[:4]
ymin = boxes[0]
xmin = boxes[1]
ymax = boxes[2]
xmax = boxes[3]
conf_score = cur_det[4]
cv2.rectangle(
image_np, (int(xmin), int(ymin)),
(int(xmax), int(ymax)), (255, 0, 0), 3
) #This is still RGB here,that's why the first element is Red
image_np_bgr = image_np[..., ::-1]
cv2.imshow('face-detection-ssd', image_np_bgr)
if cv2.waitKey(1) == 27:
break
cap.release()
cv2.destroyAllWindows()
return
def run_tf_object_detection_images(input_graph,
image_tensor,
tensor_dict,
path_to_images_dir=None):
image_list = glob.glob(path_to_images_dir + '*.jpg')
print(image_list)
with input_graph.as_default():
with tf.Session() as sess:
for cur_image_path in image_list:
while True:
image = cv2.imread(cur_image_path)
image_np = (cv2.cvtColor(image, cv2.COLOR_BGR2RGB)).astype(
np.uint8) #Convert to RGB and convert to uint8
output_dict = run_inference_for_single_image_through_ssd(
sess, image_np, image_tensor, tensor_dict)
dets = post_process_ssd_predictions(image_np,
output_dict,
threshold=0.25)
for cur_det in dets:
boxes = cur_det[:4]
ymin = boxes[0]
xmin = boxes[1]
ymax = boxes[2]
xmax = boxes[3]
conf_score = cur_det[4]
cv2.rectangle(
image_np, (int(xmin), int(ymin)),
(int(xmax), int(ymax)), (255, 0, 0), 3
) #This is still RGB here,that's why the first element is Red
image_np_bgr = image_np[..., ::-1]
cv2.imshow('face-detection-ssd', image_np_bgr)
print("Press Escape to go to next image")
if cv2.waitKey(1) == 27:
break
cv2.destroyAllWindows()
return
print("Error opening video stream or file")
while cap.isOpened():
_, image = cap.read()
image = image[..., ::-1, :]
image_display = image.copy()
initial_inference_start_time = time.time()
# Both the SSD and Facenet also uses np.uint8 and RGB images for both!
image_np = (cv2.cvtColor(image, cv2.COLOR_BGR2RGB)).astype(np.uint8)
start_time_ssd_detection = time.time()
output_dict = run_inference_for_single_image_through_ssd(sess, image_np, image_tensor, tensor_dict)
elapsed_time = time.time() - start_time_ssd_detection
# In coco dataset label map is a human being
dets = post_process_ssd_predictions(image_np, output_dict, threshold=0.5,detection_classes = [1])
print('SSD inference time cost: {}'.format(elapsed_time))
for detection_id, cur_det in enumerate(dets):
boxes = cur_det[:4]
(ymin, xmin, ymax, xmax) = (boxes[0], boxes[1],
boxes[2], boxes[3])
cv2.rectangle(image_display, (int(xmin), int(ymin)), (int(xmax), int(ymax)),
(255, 0, 0), 2)
cv2.imshow('full-face-detection-pipeline', image_display)
if cv2.waitKey(1) == 27:
break
cap.release()
cv2.destroyAllWindows()
def run_tf_object_detection_video(input_graph,
image_tensor,
tensor_dict,
PICTURE_ID,
path_to_video=None):
if path_to_video == None:
cap = cv2.VideoCapture(0)
else:
cap = cv2.VideoCapture(path_to_video)
if cap.isOpened() is False:
print("Error opening video stream or file")
with input_graph.as_default():
with tf.Session() as sess:
while cap.isOpened():
_, image = cap.read()
image = image[..., ::-1, :]
image_display = image.copy(
) # This variable is for the purpose of displaying the picture to the user
image_np = (cv2.cvtColor(image, cv2.COLOR_BGR2RGB)).astype(
np.uint8) #Convert to RGB and convert to uint8
output_dict = run_inference_for_single_image(
sess, image_np, image_tensor, tensor_dict)
dets = post_process_ssd_predictions(image_np,
output_dict,
threshold=0.5)
for cur_det in dets:
boxes = cur_det[:4]
ymin = boxes[0]
xmin = boxes[1]
ymax = boxes[2]
xmax = boxes[3]
conf_score = cur_det[4]
# This is still RGB here,that's why the first element is Red
cv2.rectangle(image_display, (int(xmin), int(ymin)),
(int(xmax), int(ymax)), (0, 255, 0), 3)
new_xmin, new_xmax, new_ymin, new_ymax = crop_ssd_prediction(
xmin, xmax, ymin, ymax, None, image_np.shape[1],
image_np.shape[0], 0.5, 0.3)
cropped_coordinates = (new_xmin, new_ymin, new_xmax,
new_ymax)
cv2.rectangle(image_display,
(int(new_xmin), int(new_ymin)),
(int(new_xmax), int(new_ymax)),
(255, 0, 255), 3)
if dets.shape[0] > 1:
cv2.putText(
image_display,
'Face is not stable or there may be more than one person in camera. Please readjust accordingly',
(0, 20),
0,
0.7, (0, 0, 255),
thickness=2)
else:
cv2.putText(image_display,
'Please make sure you have a good lighting.',
(0, 20),
0,
0.7, (0, 255, 0),
thickness=2)
cv2.imshow('face-detection-ssd', image_display)
if cv2.waitKey(1) & 0xFF == ord('y'): # save on pressing 'y'
if dets.shape[0] == 1:
cropped_img = image[
cropped_coordinates[1]:cropped_coordinates[3],
cropped_coordinates[0]:cropped_coordinates[2]]
resized_frame = cv2.resize(cropped_img, (250, 250))
resized_frame_gamma_0_5 = adjust_gamma(resized_frame,
gamma=0.5)
resized_frame_gamma_0_7 = adjust_gamma(resized_frame,
gamma=0.7)
resized_frame_gamma_1_5 = adjust_gamma(resized_frame,
gamma=1.5)
resized_frame_gamma_1_8 = adjust_gamma(resized_frame,
gamma=1.8)
cv2.imwrite(
os.path.join(TARGET_DIRECTORY,
'image_' + str(PICTURE_ID) + '.jpg'),
resized_frame)
cv2.imwrite(
os.path.join(
TARGET_DIRECTORY, 'image_' + str(PICTURE_ID) +
'_gamma' + str(0.4) + '.jpg'),
resized_frame_gamma_0_5)
cv2.imwrite(
os.path.join(
TARGET_DIRECTORY, 'image_' + str(PICTURE_ID) +
'_gamma' + str(1.8) + '.jpg'),
resized_frame_gamma_1_8)
cv2.imwrite(
os.path.join(
TARGET_DIRECTORY, 'image_' + str(PICTURE_ID) +
'_gamma' + str(0.7) + '.jpg'),
resized_frame_gamma_0_7)
cv2.imwrite(
os.path.join(
TARGET_DIRECTORY, 'image_' + str(PICTURE_ID) +
'_gamma' + str(1.5) + '.jpg'),
resized_frame_gamma_1_5)
print(
"Image ", str(PICTURE_ID),
"has been saved along with different gamma preprocessing!"
)
PICTURE_ID += 1
else:
print(
"Picture is not stable. Found two detections. Please readjust until only one face is detected!"
)
if cv2.waitKey(1) & 0xFF == ord('q'):
cv2.destroyAllWindows()
break
cap.release()
cv2.destroyAllWindows()
return
### Creating and Loading the Facenet Graph ###
images_placeholder, embeddings, phase_train_placeholder = load_tf_facenet_graph(
FACENET_MODEL_PATH)
for image_id, SOURCE_IM_PATH in enumerate(SOURCE_IM_PATH_ARRAY):
initial_inference_start_time = time.time()
image = cv2.imread(SOURCE_IM_PATH)
image_np = (cv2.cvtColor(image, cv2.COLOR_BGR2RGB)).astype(
np.uint8) # Convert to RGB and convert to uint8
start_time_ssd_detection = time.time()
output_dict = run_inference_for_single_image(
sess, image_np, image_tensor, tensor_dict)
elapsed_time = time.time() - start_time_ssd_detection
dets = post_process_ssd_predictions(image_np,
output_dict,
threshold=0.25)
print('SSD inference time cost: {}'.format(elapsed_time))
im_height = image.shape[0]
im_width = image.shape[1]
# For each of the detection boxes in dets, need to pass it to Facenet after using the facenet load data
ids = []
bbox_dict = {}
images_array = []
for detection_id, cur_det in enumerate(dets):
boxes = cur_det[:4]
(ymin, xmin, ymax, xmax) = (boxes[0], boxes[1], boxes[2],