#---------------------------------------------- #--- Author : Ahmet Ozlu #--- Mail : [email protected] #--- Date : 27th January 2018 #---------------------------------------------- # Imports import tensorflow as tf # Object detection imports from utils import backbone from api import object_counting_api input_video = "./input_images_and_videos/pedestrian_survaillance.mp4" # By default I use an "SSD with Mobilenet" model here. See the detection model zoo (https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md) for a list of other models that can be run out-of-the-box with varying speeds and accuracies. detection_graph, category_index = backbone.set_model( 'ssd_mobilenet_v1_coco_2018_01_28') fps = 30 # change it with your input video fps width = 626 # change it with your input video width height = 360 # change it with your input vide height is_color_recognition_enabled = 0 # set it to 1 for enabling the color prediction for the detected objects roi = 385 # roi line position deviation = 1 # the constant that represents the object counting area object_counting_api.cumulative_object_counting_x_axis( input_video, detection_graph, category_index, is_color_recognition_enabled, fps, width, height, roi, deviation) # counting all the objects
# Imports
import tensorflow as tf
# Object detection imports
from utils import backbone
from api import object_counting_api
if tf.__version__ < '1.10.0':
raise ImportError('Please upgrade your tensorflow installation to v1.4.* or later!')
input_video = "./input_images_and_videos/first_phase_p.mp4"
# By default I use an "SSD with Mobilenet" model here. See the detection model zoo (https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md) for a list of other models that can be run out-of-the-box with varying speeds and accuracies.
# detection_graph, category_index = backbone.set_model('ssd_mobilenet_v1_coco_2017_11_17')
detection_graph, category_index = backbone.set_model('rfcn_resnet101_coco_11_06_2017')
# detection_graph, category_index = backbone.set_model('ssd_mobilenet_v1_fpn_shared_box_predictor_640x640_coco14_sync_2018_07_03')
#object_counting_api.object_counting(input_video, detection_graph, category_index, 0) # for counting all the objects, disabled color prediction
#object_counting_api.object_counting(input_video, detection_graph, category_index, 1) # for counting all the objects, enabled color prediction
targeted_objects = "traffic light" # (for counting targeted objects) change it with your targeted objects
# targeted_objects = "my_traffic_light"
fps = 50 # change it with your input video fps
width = 1920 # change it with your input video width
height = 1080 # change it with your input vide height
is_color_recognition_enabled = 0
object_counting_api.targeted_object_counting(input_video, detection_graph, category_index, is_color_recognition_enabled, targeted_objects, fps, width, height) # targeted objects counting
# Imports
import tensorflow as tf
# Object detection imports
from utils import backbone
from api import object_counting_api
detection_graph, category_index = backbone.set_model('inference_graph',
'labelmap.pbtxt')
is_color_recognition_enabled = 0
object_counting_api.object_counting_webcam(detection_graph, category_index,
is_color_recognition_enabled)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
input_video = "Bank.mp4"
# By default I use an "SSD with Mobilenet" model here. See the detection model zoo (https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md) for a list of other models that can be run out-of-the-box with varying speeds and accuracies.
detection_graph, category_index = backbone.set_model('faster_rcnn_inception_v2_coco_2018_01_28')
#object_counting_api.object_counting(input_video, detection_graph, category_index, 0) # for counting all the objects, disabled color prediction
#object_counting_api.object_counting(input_video, detection_graph, category_index, 1) # for counting all the objects, enabled color prediction
targeted_objects = "person" # (for counting targeted objects) change it with your targeted objects
fps = 25 # change it with your input video fps
width = 1280 # change it with your input video width
height = 720 # change it with your input video height
is_color_recognition_enabled = 0
targeted_object_counting(input_video, detection_graph, category_index, is_color_recognition_enabled, targeted_objects, fps, width, height) # targeted objects counting
#object_counting_api.object_counting(input_video, detection_graph, category_index, is_color_recognition_enabled, fps, width, height) # counting all the objects
#---------------------------------------------- #--- Author : Ahmet Ozlu #--- Mail : [email protected] #--- Date : 27th January 2018 #---------------------------------------------- # Imports import tensorflow as tf # Object detection imports from utils import backbone from api import object_counting_api input_video = "./input_images_and_videos/vehicle_survaillance.mp4" # By default I use an "SSD with Mobilenet" model here. See the detection model zoo (https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md) for a list of other models that can be run out-of-the-box with varying speeds and accuracies. detection_graph, category_index = backbone.set_model( 'ssd_mobilenet_v1_coco_2018_01_28', 'mscoco_label_map.pbtxt') is_color_recognition_enabled = 0 # set it to 1 for enabling the color prediction for the detected objects roi = 200 # roi line position deviation = 3 # the constant that represents the object counting area object_counting_api.cumulative_object_counting_y_axis( input_video, detection_graph, category_index, is_color_recognition_enabled, roi, deviation) # counting all the objects
import tensorflow as tf
from utils import backbone
from api import object_counting_api
input_video = "./input_images_and_videos/object_test.mp4"
detection_graph, category_index = backbone.set_model(
'faster_rcnn_inception_v2_coco_2018_01_28', 'mscoco_label_map.pbtxt')
is_color_recognition_enabled = 0
roi = 185
deviation = 2
object_counting_api.cumulative_object_counting_y_axis(
input_video, detection_graph, category_index, is_color_recognition_enabled,
roi, deviation)
#object_counting_api.object_counting_webcam(detection_graph, category_index, is_color_recognition_enabled)
#---------------------------------------------- #--- Author : Ahmet Ozlu #--- Mail : [email protected] #--- Date : 27th January 2018 #---------------------------------------------- # Imports import tensorflow as tf # Object detection imports from utils import backbone from api import object_counting_api input_video = "input_images_and_videos/20191123_M02_1" # By default I use an "SSD with Mobilenet" model here. See the detection model zoo (https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md) for a list of other models that can be run out-of-the-box with varying speeds and accuracies. detection_graph, category_index = backbone.set_model('output_7kStepsSSD', 'parcel.pbtxt') is_color_recognition_enabled = 0 # set it to 1 for enabling the color prediction for the detected objects roi = 600 # roi line position deviation = 25 # the constant that represents the object counting area object_counting_api.couting_parcel_passed_line( input_video, detection_graph, category_index, is_color_recognition_enabled, roi, deviation, True, True) # counting all the objects
#---------------------------------------------- #--- Author : Ahmet Ozlu #--- Mail : [email protected] #--- Date : 27th January 2018 #---------------------------------------------- # Imports import tensorflow as tf # Object detection imports from utils import backbone from api import object_counting_api input_video = "./input_images_and_videos/vehicle_survaillance.mp4" # By default I use an "SSD with Mobilenet" model here. See the detection model zoo (https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md) for a list of other models that can be run out-of-the-box with varying speeds and accuracies. detection_graph, category_index = backbone.set_model('inference_graph') fps = 30 # change it with your input video fps width = 626 # change it with your input video width height = 360 # change it with your input vide height is_color_recognition_enabled = 0 # set it to 1 for enabling the color prediction for the detected objects roi = 385 # roi line position deviation = 3 # the constant that represents the object counting area object_counting_api.cumulative_object_counting_x_axis( input_video, detection_graph, category_index, is_color_recognition_enabled, fps, width, height, roi, deviation) # counting all the objects
def cumucount():
input_video = "tiv.mp4"
# input_video = "http://*****:*****@192.168.0.105:80/1"
# By default I use an "SSD with Mobilenet" model here. See the detection model zoo (https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md) for a list of other models that can be run out-of-the-box with varying speeds and accuracies.
detection_graph, category_index = backbone.set_model(
'ssd_mobilenet_v1_coco_2017_11_17')
targeted_objects = "person"
fps = 24 # change it with your input video fps
width = 640 # change it with your input video width
height = 480 # change it with your input vide height
is_color_recognition_enabled = 0 # set it to 1 for enabling the color prediction for the detected objects
roi = 200 # roi line position
deviation = 5 # the constant that represents the object counting area
total_passed_vehicle = 0
fourcc = cv2.VideoWriter_fourcc(*'XVID')
output_movie = cv2.VideoWriter('the_output.avi', fourcc, fps,
(width, height))
# input video
cap = cv2.VideoCapture(input_video)
cam_resolution = RESOLUTION_VGA
model_detector = FaceDetectorModels.HAARCASCADE
model_recognizer = FaceEncoderModels.LBPH
try:
# Initialize face detection
face_detector = FaceDetector(model=model_detector,
path=INPUT_DIR_MODEL_DETECTION)
# Initialize face recognizer
face_encoder = FaceEncoder(model=model_recognizer,
path=INPUT_DIR_MODEL_ENCODING,
path_training=INPUT_DIR_MODEL_TRAINING,
training=False)
except:
face_encoder = None
print("Warning, check if models and trained dataset models exists!")
face_id, confidence = (None, 0)
total_passed_vehicle = 0
counting_mode = "..."
width_heigh_taken = True
with detection_graph.as_default():
with tf.Session(graph=detection_graph) as sess:
# Definite input and output Tensors for detection_graph
image_tensor = detection_graph.get_tensor_by_name('image_tensor:0')
# Each box represents a part of the image where a particular object was detected.
detection_boxes = detection_graph.get_tensor_by_name(
'detection_boxes:0')
# Each score represent how level of confidence for each of the objects.
# Score is shown on the result image, together with the class label.
detection_scores = detection_graph.get_tensor_by_name(
'detection_scores:0')
detection_classes = detection_graph.get_tensor_by_name(
'detection_classes:0')
num_detections = detection_graph.get_tensor_by_name(
'num_detections:0')
# for all the frames that are extracted from input video
while (cap.isOpened()):
ret, frame = cap.read()
if not ret:
print("end of the video file...")
break
input_frame = frame
# Detect and identify faces in the frame
faces = face_detector.detect(input_frame)
for (index, face) in enumerate(faces):
(x, y, w, h) = face
# Identify face based on trained dataset (note: should run facial_recognition_training.py)
if face_encoder is not None:
face_id, confidence = face_encoder.identify(
input_frame, (x, y, w, h))
# Set text and bounding box on face
label_face(input_frame, (x, y, w, h), face_id, confidence)
# Process 1 face only
#break
# Expand dimensions since the model expects images to have shape: [1, None, None, 3]
image_np_expanded = np.expand_dims(input_frame, axis=0)
# Actual detection
(boxes, scores, classes,
num) = sess.run([
detection_boxes, detection_scores, detection_classes,
num_detections
],
feed_dict={image_tensor: image_np_expanded})
# insert information text to video frame
font = cv2.FONT_HERSHEY_SIMPLEX
# Visualization of the results of a detection.
counter, csv_line, counting_mode = vis_util.visualize_boxes_and_labels_on_image_array_y_axis(
cap.get(1),
input_frame,
2,
is_color_recognition_enabled,
np.squeeze(boxes),
np.squeeze(classes).astype(np.int32),
np.squeeze(scores),
category_index,
targeted_objects="person",
y_reference=roi,
deviation=deviation,
use_normalized_coordinates=True,
line_thickness=4)
# when the vehicle passed over line and counted, make the color of ROI line green
if counter == 1:
cv2.line(input_frame, (0, roi), (width, roi), (0, 0xFF, 0),
5)
else:
cv2.line(input_frame, (0, roi), (width, roi), (0, 0, 0xFF),
5)
total_passed_vehicle = total_passed_vehicle + counter
# insert information text to video frame
font = cv2.FONT_HERSHEY_SIMPLEX
cv2.putText(
input_frame,
'Detected: ' + str(total_passed_vehicle),
(10, 35),
font,
0.8,
(0, 0xFF, 0xFF),
2,
cv2.FONT_HERSHEY_SIMPLEX,
)
cv2.putText(
input_frame,
'ROI Line',
(545, roi - 10),
font,
0.6,
(0, 0, 0xFF),
2,
cv2.LINE_AA,
)
output_movie.write(input_frame)
#print ("writing frame")
#cv2.imshow('object counting',input_frame)
#if cv2.waitKey(1) & 0xFF == ord('q'):
#break
# Display updated frame to web app
yield (b'--frame\r\nContent-Type: image/jpeg\r\n\r\n' +
cv2.imencode('.jpg', frame)[1].tobytes() + b'\r\n\r\n')
cap.release()
cv2.destroyAllWindows()
import warnings
warnings.filterwarnings('ignore')
print("Trying to import tf")
import tensorflow as tf
print("imported tf")
sys.path.append("..")
from utils import label_map_util
from utils import visualization_utils as vis_util
from utils import backbone
import json
from livepredictor import cumulative_object_counting_x_axis
input_video = "rtsp://*****:*****@46.52.182.186:37778/cam/realmonitor?channel=2&subtype=0"
detection_graph, category_index = backbone.set_model(
'oldtestped_inference_graph', 'object-detection.pbtxt')
is_color_recognition_enabled = 0 # set it to 1 for enabling the color prediction for the detected objects
label = "pedestrian"
#for door
x = 820
y = 66
w = 468
h = 800
'''
x= 1500
y = 22
w = 468
