You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 

294 lines
11 KiB

#!/usr/bin/python3
import pickle
import threading
import sys
import cv2
import os
import numpy as np
from utils import label_map_util
from utils import visualization_utils as vis_util
if sys.platform == "win32":
import tensorflow as tf
from distutils.version import StrictVersion
if StrictVersion(tf.__version__) < StrictVersion('1.12.0'):
raise ImportError('Please upgrade your TensorFlow installation to v1.12.*.')
else:
import psutil
import json
import base64
from PIL import Image
from io import BytesIO
switch = 1
import socket
# This is needed since the notebook is stored in the object_detection folder.
sys.path.append("..")
import time
from object_detection.utils import ops as utils_ops
# What model to download.
encode_param = [int(cv2.IMWRITE_JPEG_QUALITY), 90]
#MODEL_NAME = 'ssd_mobilenet_v1_coco_2017_11_17' #not even worth trying
#MODEL_NAME = "ssd_inception_v2_coco_2017_11_17" # not bad and fast
MODEL_NAME = "rfcn_resnet101_coco_2018_01_28" # WORKS BEST BUT takes 4 times longer per image
#MODEL_NAME = "faster_rcnn_resnet101_coco_11_06_2017" # too slow
#MODEL_NAME = "ssd_resnet101_v1_fpn_shared_box_predictor_oid_512x512_sync_2019_01_20"
MODEL_FILE = MODEL_NAME + '.tar.gz'
DOWNLOAD_BASE = 'http://download.tensorflow.org/models/object_detection/'
# Path to frozen detection graph. This is the actual model that is used for the object detection.
PATH_TO_FROZEN_GRAPH = MODEL_NAME + '/frozen_inference_graph.pb'
# List of the strings that is used to add correct label for each box.
PATH_TO_LABELS = os.path.join('object_detection/data', 'mscoco_label_map.pbtxt')
category_index = label_map_util.create_category_index_from_labelmap(PATH_TO_LABELS, use_display_name=True)
if sys.platform == "win32":
detection_graph = tf.Graph()
with detection_graph.as_default():
od_graph_def = tf.GraphDef()
with tf.gfile.GFile(PATH_TO_FROZEN_GRAPH, 'rb') as fid:
serialized_graph = fid.read()
od_graph_def.ParseFromString(serialized_graph)
tf.import_graph_def(od_graph_def, name='')
def load_image_into_numpy_array(image):
(im_width, im_height) = image.size
return np.array(image.getdata()).reshape(
(im_height, im_width, 3)).astype(np.uint8)
# For the sake of simplicity we will use only 2 images:
# image1.jpg
# image2.jpg
# If you want to test the code with your images, just add path to the images to the TEST_IMAGE_PATHS.
PATH_TO_TEST_IMAGES_DIR = 'object_detection/test_images'
TEST_IMAGE_PATHS = [ os.path.join(PATH_TO_TEST_IMAGES_DIR, 'image{}.jpg'.format(i)) for i in range(3, 6) ]
# Size, in inches, of the output images.
sess = 0
data = {"gpu_temp":"10C","gpu_load":"15%","cpu_temp":"47C","cpu_load":"15%","mem_temp":"NaN","mem_load":"17%","fan_speed":"10000RPM"}
def get_temps():
global data
if not sys.platform == "win32":
temps = psutil.sensors_temperatures()
data["cpu_temp"] = str(int(temps["dell_smm"][0][1]))+"°C"
data["cpu_load"] = str(psutil.cpu_percent())+"%"
data["mem_load"] = str(dict(psutil.virtual_memory()._asdict())["percent"])+"%"
data["fan_speed"] = str(psutil.sensors_fans()["dell_smm"][0][1])+"RPM"
def run_inference_for_single_image(image, graph):
global switch
global sess
with graph.as_default():
if(switch):
sess = tf.Session()
switch = 0
# Get handles to input and output tensors
ops = tf.get_default_graph().get_operations()
all_tensor_names = {output.name for op in ops for output in op.outputs}
tensor_dict = {}
for key in [
'num_detections', 'detection_boxes', 'detection_scores',
'detection_classes', 'detection_masks'
]:
tensor_name = key + ':0'
if tensor_name in all_tensor_names:
tensor_dict[key] = tf.get_default_graph().get_tensor_by_name(
tensor_name)
if 'detection_masks' in tensor_dict:
# The following processing is only for single image
detection_boxes = tf.squeeze(tensor_dict['detection_boxes'], [0])
detection_masks = tf.squeeze(tensor_dict['detection_masks'], [0])
# Reframe is required to translate mask from box coordinates to image coordinates and fit the image size.
real_num_detection = tf.cast(tensor_dict['num_detections'][0], tf.int32)
detection_boxes = tf.slice(detection_boxes, [0, 0], [real_num_detection, -1])
detection_masks = tf.slice(detection_masks, [0, 0, 0], [real_num_detection, -1, -1])
detection_masks_reframed = utils_ops.reframe_box_masks_to_image_masks(
detection_masks, detection_boxes, image.shape[1], image.shape[2])
detection_masks_reframed = tf.cast(
tf.greater(detection_masks_reframed, 0.5), tf.uint8)
# Follow the convention by adding back the batch dimension
tensor_dict['detection_masks'] = tf.expand_dims(
detection_masks_reframed, 0)
image_tensor = tf.get_default_graph().get_tensor_by_name('image_tensor:0')
# Run inference
output_dict = sess.run(tensor_dict,
feed_dict={image_tensor: image})
# all outputs are float32 numpy arrays, so convert types as appropriate
output_dict['num_detections'] = int(output_dict['num_detections'][0])
output_dict['detection_classes'] = output_dict[
'detection_classes'][0].astype(np.int64)
output_dict['detection_boxes'] = output_dict['detection_boxes'][0]
output_dict['detection_scores'] = output_dict['detection_scores'][0]
if 'detection_masks' in output_dict:
output_dict['detection_masks'] = output_dict['detection_masks'][0]
return output_dict
kill = True
def listener(port=8385):
serversocket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
serversocket.bind((socket.gethostname(), port))
serversocket.listen(5)
while kill:
serversocket.accept()
print('Bye!')
cut = [115, 100, 400, 150]
cut_send = [0, 0, 0, 0]
img_counter = 0
socket_switch = True
thread = threading.Thread(target=listener)
thread.start()
if sys.platform == "win32":
with detection_graph.as_default():
sess = tf.Session()
cam = cv2.VideoCapture(0)
else:
cam = cv2.VideoCapture('debug_data/amb_1.mp4')
with open("debug_data/frame_data.pkl","rb") as pkl_file:
frame_data = pickle.load(pkl_file)
switch = 0
get_temps()
amb_center = {'x': (400 + 550)/2, 'y': (115+215)/2}
reps = -1
reps_vid = 0
while 1:
ret,image = cam.read()
reps_vid += 1
if not sys.platform == "win32" and not reps_vid % 2 == 0:
continue
reps += 1
try: # Kavşak
t1 = time.time()
image_np = image
image_np_expanded = np.expand_dims(image_np, axis=0)
if sys.platform == "win32":
output_dict = run_inference_for_single_image(image_np_expanded, detection_graph)
else:
output_dict = frame_data[reps]
height, width, channels = image_np.shape
out_dict = {'detection_boxes': [], 'detection_classes': [], 'detection_scores': []}
for i in output_dict['detection_classes']:
cont = False
if i in [3, 6, 8]: # Car, bus, truck
index = np.where(output_dict['detection_classes'] == i)[0][0]
score = output_dict['detection_scores'][index]
if score > 0.3:
if not any((output_dict['detection_boxes'][index] == b).all() for b in out_dict['detection_boxes']):
avg_x = (output_dict['detection_boxes'][index][0] + output_dict['detection_boxes'][index][2])/2
avg_y = (output_dict['detection_boxes'][index][1] + output_dict['detection_boxes'][index][3])/2
for box in out_dict['detection_boxes']:
avg_box_x = (box[0] + box[2])/2
avg_box_y = (box[1] + box[3])/2
if abs(avg_x-avg_box_x) < 0.1 and abs(avg_y-avg_box_y) < 0.1:
cont = True
continue
out_dict['detection_classes'].append(i)
out_dict['detection_boxes'].append(output_dict['detection_boxes'][index])
out_dict['detection_scores'].append(output_dict['detection_scores'][index])
out_dict['detection_classes'] = np.array(out_dict['detection_classes'])
out_dict['detection_boxes'] = np.array(out_dict['detection_boxes'])
out_dict['detection_scores'] = np.array(out_dict['detection_scores'])
print(len(out_dict['detection_classes']), ' cars.')
vis_util.visualize_boxes_and_labels_on_image_array(
image_np,
out_dict['detection_boxes'],
out_dict['detection_classes'],
out_dict['detection_scores'],
category_index,
instance_masks=out_dict.get('detection_masks'),
use_normalized_coordinates=True,
line_thickness=8,
min_score_thresh=0.3
)
cv2.imshow('frame', image_np)
ex_c = [27, ord("q"), ord("Q")]
if cv2.waitKey(1) & 0xFF in ex_c:
break
t2 = time.time()
print("time taken for {}".format(t2-t1))
if not sys.platform == "win32" and t2-t1 < 0.1:
time.sleep(0.1-(t2-t1))
send_image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
if socket_switch:
try:
client_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
client_socket.settimeout(0.1)
client_socket.connect(('127.0.0.1', 8485))
connection = client_socket.makefile('wb')
socket_switch = False
except:
socket_switch = True
continue
try:
crop_img = send_image.copy(order='C')
crop_img = Image.fromarray(crop_img, "RGB")
buffered = BytesIO()
crop_img.save(buffered, format="JPEG")
img = base64.b64encode(buffered.getvalue()).decode("ascii")
lens = [len(send_image), 0, len(send_image[0])]
for i in range(0,len(cut), 2):
if cut[i] < 0:
cut_send[i] = lens[i] + cut[i]
cut_send[i+1] = abs(cut[i])-abs(cut[i+1])
client_socket.sendall(json.dumps({"image_full":img,"image_sizes":{"x":cut_send[2],"y":cut_send[0],"width":cut_send[3],"height":cut_send[1]},"load":data}).encode('gbk')+b"\n")
img_counter += 1
except:
socket_switch = True
if img_counter % 10 == 0:
get_temps()
except Exception as e:
if hasattr(e, 'message'):
print(e.message)
else:
print(e)
break
if not socket_switch:
client_socket.sendall(b"Bye\n")
cam.release()
cv2.destroyAllWindows()
cam.release()
kill = False
thread.join()