TensorFlow OpenCV Slack Flack
TensorFlow OpenCV Slack Flack
main.py : 웹 메인 서버
#!/usr/bin/env python
#
# Project: Streaming Tensorflow image with Flask
# Author: jframework@gmail.com
# Date: 2020/05/11
# Website: http://www.joang.com
# Description:
# Publishing a video from room , Tensorflow and OpenCV and GPIO, Flask
# Usage:
# 1. Install Python dependencies: tensorflow, opencv, gpio, cv2, flask. (wish that pip install works like a charm)
# 2. Run "python3 main.py".
# 3. Navigate the browser to the local webpage. http://web.joang.com:8081/
#
#
from flask import Flask, render_template, Response, request
from camera import VideoCamera
from gpioControl import GpioControl
import os
app = Flask(__name__)
def shutdown_server():
func = request.environ.get('werkzeug.server.shutdown')
if func is None:
raise RuntimeError('Not running with the kkoRack Server')
func()
ONAIR = True
gpio = GpioControl()
@app.route('/')
def index():
temp = os.popen("vcgencmd measure_temp").readline()
return render_template('index.html', temperature=(temp.replace("temp=","")))
## Video ##
def gen(camera):
global ONAIR
while(ONAIR):
frame = camera.get_frame()
yield (b'--frame\r\n'
b'Content-Type: image/jpeg\r\n\r\n' + frame + b'\r\n\r\n')
@app.route('/video_feed')
def video_feed():
return Response(gen(VideoCamera()),
mimetype='multipart/x-mixed-replace; boundary=frame')
## Init ##
@app.route('/init')
def video_init():
print("video_init")
reString = gpio.initMotorPosition()
return "Camera position init : " + reString
## camera movement ##
@app.route('/right')
def video_right():
print("video_right")
reString = gpio.moveRight()
return "Camera Move right : " + reString
@app.route('/left')
def video_left():
print("video_left")
reString = gpio.moveLeft()
return "Camera Move left : " + reString
@app.route('/up')
def video_up():
print("video_up")
reString = gpio.moveUp()
return "Camera Move up : " + reString
@app.route('/down')
def video_down():
print("video_down")
reString = gpio.moveDown()
return "Camera Move down : " + reString
## gpio on/off ##
@app.route('/clicklight')
def clicklight():
reString = gpio.click()
return "Light Click"
## System ##
@app.route('/stop')
def video_stop():
global ONAIR
ONAIR = False
print("video_stop")
return render_template('index.html')
@app.route('/start')
def video_start():
global ONAIR
ONAIR = True
return render_template('index.html')
@app.route('/shutdown')
def shutdown():
gpio.cleanUp()
shutdown_server()
return 'Server shutting down...'
@app.route('/temperature')
def temperature():
temp = os.popen("vcgencmd measure_temp").readline()
return (temp.replace("temp=",""))
if __name__ == '__main__':
app.run(host='0.0.0.0', port='8081', debug=True)camera.py : 카메라 기능
import cv2
from objectDetector import ObjectDetector
# Set up camera constants
IM_WIDTH = 640
IM_HEIGHT = 480
# Initialize frame rate calculation
frame_rate_calc = 1
freq = cv2.getTickFrequency()
font = cv2.FONT_HERSHEY_SIMPLEX
objDetector = ObjectDetector()
class VideoCamera(object):
def __init__(self):
# Using OpenCV to capture from device 0. If you have trouble capturing
# from a webcam, comment the line below out and use a video file
# instead.
self.video = cv2.VideoCapture(0)
# If you decide to use video.mp4, you must have this file in the folder
# as the main.py.
# self.video = cv2.VideoCapture('video.mp4')
print( "width: {}, height : {}".format(self.video.get(3), self.video.get(4) ) )
ret = self.video.set(3,IM_WIDTH)
ret = self.video.set(4,IM_HEIGHT)
def __del__(self):
self.video.release()
def get_frame(self):
global frame_rate_calc
t1 = cv2.getTickCount()
success, frame = self.video.read()
# We are using Motion JPEG, but OpenCV defaults to capture raw images,
# so we must encode it into JPEG in order to correctly display the
# video stream.
frame = objDetector.objectDetector(frame)
# Draw FPS
cv2.putText(frame,"FPS: {0:.2f}".format(frame_rate_calc),(30,50),font,1,(255,255,0),2,cv2.LINE_AA)
# FPS calculation
t2 = cv2.getTickCount()
time1 = (t2-t1)/freq
frame_rate_calc = 1/time1
ret, jpeg = cv2.imencode('.jpg', frame)
return jpeg.tobytes()objectDetector.py : 이미지 식별
#!/usr/bin/python
# -*- coding: UTF-8 -*-
# Import packages
import os
import cv2
import numpy as np
import tensorflow as tf
from time import sleep
import sys
from gpioControl import GpioControl
from slackmsg import SlackMsg
# Set up camera constants
IM_WIDTH = 640
IM_HEIGHT = 480
# This is needed since the working directory is the object_detection folder.
sys.path.append('/home/pi/tensorflow1/models/research')
sys.path.append('/home/pi/tensorflow1/models/research/object_detection')
# Import utilites
from utils import label_map_util
from utils import visualization_utils as vis_util
# Grab path to current working directory
#CWD_PATH = os.getcwd() + "/tensorflow1/models/research/object_detection"
CWD_PATH = "/home/pi/tensorflow1/models/research/object_detection"
#IMG_PATH = os.getcwd() + "/Pictures"
IMG_PATH = "/home/pi/Pictures"
print(" \n ###################################")
print(" Base Path %s" % CWD_PATH)
print(" Image Path %s" % IMG_PATH)
print(" ###################################")
#### Initialize TensorFlow model ####
# Name of the directory containing the object detection module we're using
MODEL_NAME = 'ssdlite_mobilenet_v2_coco_2018_05_09'
# Path to frozen detection graph .pb file, which contains the model that is used
# for object detection.
PATH_TO_CKPT = os.path.join(CWD_PATH,MODEL_NAME,'frozen_inference_graph.pb')
# Path to label map file
PATH_TO_LABELS = os.path.join(CWD_PATH,'data','mscoco_label_map.pbtxt')
# Number of classes the object detector can identify
NUM_CLASSES = 80
print(" \n ###################################")
print(" Load the label map ")
print(" ###################################")
## Load the label map.
# Label maps map indices to category names, so that when the convolution
# network predicts `5`, we know that this corresponds to `airplane`.
# Here we use internal utility functions, but anything that returns a
# dictionary mapping integers to appropriate string labels would be fine
label_map = label_map_util.load_labelmap(PATH_TO_LABELS)
categories = label_map_util.convert_label_map_to_categories(label_map, max_num_classes=NUM_CLASSES, use_display_name=True)
category_index = label_map_util.create_category_index(categories)
# Load the Tensorflow model into memory.
detection_graph = tf.Graph()
with detection_graph.as_default():
od_graph_def = tf.compat.v1.GraphDef()
with tf.io.gfile.GFile(PATH_TO_CKPT, 'rb') as fid:
serialized_graph = fid.read()
od_graph_def.ParseFromString(serialized_graph)
tf.import_graph_def(od_graph_def, name='')
sess = tf.compat.v1.Session(graph=detection_graph)
print(" \n ###################################")
print(" Define input and output tensors (i.e. data) for the object detection classifier ")
print(" ###################################")
# Input tensor is the image
image_tensor = detection_graph.get_tensor_by_name('image_tensor:0')
# Output tensors are the detection boxes, scores, and classes
# 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 represents level of confidence for each of the objects.
# The 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')
# Number of objects detected
num_detections = detection_graph.get_tensor_by_name('num_detections:0')
print(" \n ###################################")
print(" Initialize other parameters ")
print(" ###################################")
# Initialize frame rate calculation
frame_rate_calc = 1
freq = cv2.getTickFrequency()
font = cv2.FONT_HERSHEY_SIMPLEX
# Initialize control variables used for pet detector
detected_inside = False
detected_outside = False
inside_counter = 0
outside_counter = 0
ObjX = 0
ObjY = 0
pause = 0
pause_counter = 0
gpio = GpioControl()
slackmsg = SlackMsg()
class ObjectDetector(object):
def objectDetector(self, frame):
# Use globals for the control variables so they retain their value after function exits
global detected_inside, detected_outside
global inside_counter, outside_counter
global pause, pause_counter
frame_expanded = np.expand_dims(frame, axis=0)
# Perform the actual detection by running the model with the image as input
(boxes, scores, classes, num) = sess.run(
[detection_boxes, detection_scores, detection_classes, num_detections],
feed_dict={image_tensor: frame_expanded})
# Draw the results of the detection (aka 'visulaize the results')
vis_util.visualize_boxes_and_labels_on_image_array(
frame,
np.squeeze(boxes),
np.squeeze(classes).astype(np.int32),
np.squeeze(scores),
category_index,
use_normalized_coordinates=True,
line_thickness=8,
min_score_thresh=0.40)
# Draw boxes defining "outside" locations.
TL_outside = (int(IM_WIDTH*0.6),int(IM_HEIGHT*0.25))
BR_outside = (int(IM_WIDTH*0.85),int(IM_HEIGHT*.85))
cv2.rectangle(frame,TL_outside,BR_outside,(255,20,20),3)
cv2.putText(frame,"Outside room",(TL_outside[0]+10,TL_outside[1]-10),font,1,(255,20,255),3,cv2.LINE_AA)
# Check the class of the top detected object by looking at classes[0][0].
# If the top detected object is a person (73)
# boxes ( xmin, xmax, ymin, ymax )
threshold = 0.8
for index, value in enumerate(classes[0]):
ymin = boxes[0][index][0] * IM_HEIGHT
xmin = boxes[0][index][1] * IM_WIDTH
ymax = boxes[0][index][2] * IM_HEIGHT
xmax = boxes[0][index][3] * IM_WIDTH
#print(' =1= %s' % category_index.get(value) )
if category_index.get(value) != None:
personclassname = (category_index.get(value)).get('name')
widthvalue = int((xmax - xmin) / 2) # width 길이
heightvalue = int((ymax - ymin) / 2) # height 길이
if( scores[0, index] != 0.0 ): print('> Score = %s, Object = %s , pause = %s' % (scores[0, index], personclassname, pause) )
if scores[0, index] > threshold and personclassname == 'person' and pause == 0:
print('> Detected %s' % personclassname )
ObjX = int(((boxes[0][0][1]+boxes[0][0][3])/2)*IM_WIDTH)
ObjY = int(((boxes[0][0][0]+boxes[0][0][2])/2)*IM_HEIGHT)
# Draw a circle at center of object
cv2.circle(frame,(ObjX,ObjY), 5, (75,13,180), -1)
# If object is in outside box, increment outside counter variable
if ((ObjX > TL_outside[0]) and (ObjX < BR_outside[0]) and (ObjY > TL_outside[1]) and (ObjY < BR_outside[1])):
outside_counter = outside_counter + 1
else :
inside_counter = inside_counter + 1
# If pet has been detected inside for more than 10 frames, set detected_inside flag
# and send a text to the phone.
if inside_counter > 10:
detected_inside = True
cv2.imwrite(IMG_PATH+'/inside_' + str(ObjX) + '_' + str(ObjY) + '_counter.jpg', frame)
captureImg=os.path.join(IMG_PATH, 'inside_' + str(ObjX) + '_' + str(ObjY) + '_counter.jpg')
response = slackmsg.uploadImage("Inside Photo", captureImg)
captureImg=response['file']['permalink']
slackmsg.sendMsg("Inside", captureImg)
# Set move to detected object
gpio.move_to_position(ObjX, ObjY)
inside_counter = 0
outside_counter = 0
ObjX = 0
ObjY = 0
# Pause pet detection by setting "pause" flag
pause = 1
# If pet has been detected outside for more than 10 frames, set detected_outside flag
# and send a text to the phone.
if outside_counter > 10:
detected_outside = True
cv2.imwrite(IMG_PATH+'/outside_' + str(ObjX) + '_' + str(ObjY) + '_counter.jpg', frame)
captureImg=os.path.join(IMG_PATH, 'outside_' + str(ObjX) + '_' + str(ObjY) + '_counter.jpg')
response = slackmsg.uploadImage("Outside Photo", captureImg)
captureImg=response['file']['permalink']
slackmsg.sendMsg("Outside", captureImg)
# Set move to detected object
gpio.move_to_position(ObjX, ObjY)
inside_counter = 0
outside_counter = 0
ObjX = 0
ObjY = 0
# Pause pet detection by setting "pause" flag
pause = 1
# If pause flag is set, draw message on screen.
if pause == 1:
if detected_inside == True:
cv2.putText(frame,'Inside!',(int(IM_WIDTH*.1),int(IM_HEIGHT*.5)),font,1,(0,0,0),7,cv2.LINE_AA)
cv2.putText(frame,'Inside!',(int(IM_WIDTH*.1),int(IM_HEIGHT*.5)),font,1,(95,176,23),5,cv2.LINE_AA)
if detected_outside == True:
cv2.putText(frame,'Outside!',(int(IM_WIDTH*.1),int(IM_HEIGHT*.5)),font,1,(0,0,0),7,cv2.LINE_AA)
cv2.putText(frame,'Outside!',(int(IM_WIDTH*.1),int(IM_HEIGHT*.5)),font,1,(95,176,23),5,cv2.LINE_AA)
# Increment pause counter until it reaches 30 (for a framerate of 1.5 FPS, this is about 20 seconds),
# then unpause the application (set pause flag to 0).
pause_counter = pause_counter + 1
if pause_counter > 30:
pause = 0
pause_counter = 0
detected_inside = False
detected_outside = False
# Draw counter info
cv2.putText(frame,'Detection counter: ' + str(max(inside_counter,outside_counter)),(10,100),font,0.5,(255,255,0),1,cv2.LINE_AA)
cv2.putText(frame,'Pause counter: ' + str(pause_counter),(10,150),font,0.5,(255,255,0),1,cv2.LINE_AA)
return framegpioControl.py : 등 켜기 등 릴ㄹ레이 모듈 , 카메라 위치 조정 스텝 모터 조정
import RPi.GPIO as GPIO
from time import sleep
GPIO.setmode(GPIO.BOARD)
GPIO.setup(12, GPIO.OUT, initial=1)
GPIO.setup(18, GPIO.OUT, initial=1)
GPIO.setup(11, GPIO.OUT, initial=1) # light
p1 = GPIO.PWM(12, 50) # 50 Hz
p2 = GPIO.PWM(18, 50) # 50
p1.start(0)
p2.start(0)
p1.ChangeDutyCycle(0)
p2.ChangeDutyCycle(0)
verticalVal = 6.5
horizontalVal = 6.5
cameraPositionX = 6.5
cameraPositionY = 6.5
# Set up camera constants
IM_WIDTH = 640
IM_HEIGHT = 480
class GpioControl(object):
def __init__(self):
global verticalVal
global horizontalVal
global p1
global p2
p1.ChangeDutyCycle(6.5)
p2.ChangeDutyCycle(6.5)
sleep(0.1)
p1.ChangeDutyCycle(0)
p2.ChangeDutyCycle(0)
verticalVal = 6.5
horizontalVal = 6.5
print("> Init Vert=" + str(verticalVal) + ",Hort=" + str(horizontalVal))
def click(self):
GPIO.output(11, GPIO.LOW)
sleep(0.5)
GPIO.output(11, GPIO.HIGH)
sleep(1)
def __del__(self):
global p1
global p2
p1.stop()
p2.stop()
print(" GPIO.__del__() ")
GPIO.cleanup()
def cleanUp(self):
global p1
global p2
p1.stop()
p2.stop()
print(" GPIO.cleanUp() ")
GPIO.cleanup()
sleep(2)
def initMotorPosition(self):
# Init
global verticalVal
global horizontalVal
global p1
global p2
p1.ChangeDutyCycle(6.5)
p2.ChangeDutyCycle(6.5)
sleep(0.1)
p1.ChangeDutyCycle(0)
p2.ChangeDutyCycle(0)
verticalVal = 6.5
horizontalVal = 6.5
print("> Init Vert=" + str(verticalVal) + ",Hort=" + str(horizontalVal))
return "Vert=" + str(verticalVal) + ",Hort=" + str(horizontalVal)
def moveUp(self):
global verticalVal
global horizontalVal
global p2
verticalVal = round(verticalVal+0.2, 1)
p2.ChangeDutyCycle(verticalVal)
print("> UP Vert=" + str(verticalVal) + ",Hort=" + str(horizontalVal))
sleep(0.1)
p2.ChangeDutyCycle(0)
return "Vert=" + str(verticalVal) + ",Hort=" + str(horizontalVal)
def moveDown(self):
global verticalVal
global horizontalVal
global p2
verticalVal = round(verticalVal-0.2, 1)
p2.ChangeDutyCycle(verticalVal)
print("> Down Vert=" + str(verticalVal) + ",Hort=" + str(horizontalVal))
sleep(0.1)
p2.ChangeDutyCycle(0)
return "Vert=" + str(verticalVal) + ",Hort=" + str(horizontalVal)
def moveRight(self):
global verticalVal
global horizontalVal
global p1
horizontalVal = round(horizontalVal+0.2, 1)
p1.ChangeDutyCycle(horizontalVal)
print("> Right Vert=" + str(verticalVal) + ",Hort=" + str(horizontalVal))
sleep(0.1)
p1.ChangeDutyCycle(0)
return "Vert=" + str(verticalVal) + ",Hort=" + str(horizontalVal)
def moveLeft(self):
global verticalVal
global horizontalVal
global p1
horizontalVal = round(horizontalVal-0.2, 1)
p1.ChangeDutyCycle(horizontalVal)
print("> Left Vert=" + str(verticalVal) + ",Hort=" + str(horizontalVal))
sleep(0.1)
p1.ChangeDutyCycle(0)
return "Vert=" + str(verticalVal) + ",Hort=" + str(horizontalVal)
def move_to_position(self,ObjX, ObjY):
global cameraPositionX
global cameraPositionY
global p1
global p2
print(" >> Move to location x=%s, y=%s" % (ObjX, ObjY))
moveLoop = True
movX = int(IM_WIDTH/2)-ObjX
movY = int(IM_HEIGHT/2)-ObjY
print(" >> Center location movX=%s, movY=%s" % (movX, movY))
xx = 1
xy = 0
yx = 1
yy = 0
while(moveLoop):
if( xy < abs(movX) ):
p1.ChangeDutyCycle(cameraPositionX)
sleep(0.1)
p1.ChangeDutyCycle(0)
print("xx=" + str(xx) + ", xy="+ str(xy) +" cameraPositionX=" +str(round(cameraPositionX,1)))
xy = xx*xx * 12
xx = xx + 1
if(movX > 0): cameraPositionX = cameraPositionX - 0.2
else: cameraPositionX = cameraPositionX + 0.2
if( yy < abs(movY) ):
p2.ChangeDutyCycle(cameraPositionY)
sleep(0.1)
p2.ChangeDutyCycle(0)
print("yx=" + str(yx) + ", yy="+ str(yy) +" cameraPositionY=" +str(round(cameraPositionY,1)))
yy = yx*yx * 12
yx = yx + 1
if(movY > 0): cameraPositionY = cameraPositionY + 0.2
else: cameraPositionY = cameraPositionY - 0.2
elif( xy >= movX and yy >= movY):
print(" >> Center location movX="+str(movX)+", movY=" + str(movY))
print(" >> Position location xy="+str(xy)
+", yy="+str(yy)
+" cameraPositionX="+str(round(cameraPositionX,1))+
" cameraPositionY="+str(round(cameraPositionY,1))+" .. ")
moveLoop = False