imageRecognition.py

#! /usr/bin/python
# -*- coding: utf-8 -*-

#
import base64
import httplib2
import urllib2, json
import numpy as np
import cv2
import sys
import time

from googleapiclient import discovery
from oauth2client.client import GoogleCredentials
from picamera.array import PiRGBArray
from picamera import PiCamera

#
# The url template to retrieve the discovery document for trusted testers.
DISCOVERY_URL='https://{api}.googleapis.com/$discovery/rest?version={apiVersion}'
MAX_NUM_IMAGE_DETECTION = 5

inputFlag = False
filename = 'panasonic2.jpg'

def getImage():
     # initialize the camera and grab a reference to the raw camera capture
     camera = PiCamera()
     camera.resolution = (640, 480)
     camera.framerate = 32
     rawCapture = PiRGBArray(camera, size=(640, 480))

     # print("setting is end!")

     # allow the camera to warmup
     # time.sleep(0.1)
     dst = None
     # capture frames from the camera
     for frame in camera.capture_continuous(rawCapture, format="bgr", use_video_port=True):
	# grab the raw NumPy array representing the image, then initialize the timestamp
	# and occupied/unoccupied text
	image = frame.array

	mtx =np.array([[100, 0, 320], [0, 100, 240], [0, 0, 1]], dtype=np.float32)  
	dist = np.array([-0.009, 0.0, 0.0, 0.0], dtype=np.float32)
	h,  w = image.shape[:2]
	newcameramtx, roi=cv2.getOptimalNewCameraMatrix(mtx,dist,(w,h),1,(w,h))
	# undistort
	dst = cv2.undistort(image, mtx, dist, None, newcameramtx)
       break

     return dst

#
def imgeDeterminParamCheck(mode):
    if mode == 'LOGO_DETECTION':
        return True
    elif mode == 'LABEL_DETECTION':
        return True
    elif mode == 'FACE_DETECTION':
        return True
    elif mode == 'TEXT_DETECTION':
        return True
    else: 
        return False

#
def imageDetermin(mode, compStr, rect):
    del rect[:]
    rect.append(-1)
    rect.append(-1)
    rect.append(-1)
    rect.append(-1)
    
    if imgeDeterminParamCheck(mode) == False:
        return '-1'

    #
    """Run a label request on a single image"""
    if inputFlag == True:
        img = cv2.imread(filename, 1)
    else:
        img = getImage()
    if img is None:
        return '-1'

    img = cv2.resize(img, (int(480.0/img.shape[0]*img.shape[1]), 480) )

    #
    credentials         = GoogleCredentials.get_application_default()
    service             = discovery.build('vision', 'v1', credentials=credentials,
                              discoveryServiceUrl=DISCOVERY_URL)
    result, imgencode   = cv2.imencode('.jpg', img)
    data                = np.array(imgencode)
    image_content       = base64.b64encode(data.tostring())
    service_request     = service.images().annotate(body={
        'requests': [{
            'image': {
                'content': image_content.decode('UTF-8')
            },
            'features': [{
                'type': mode,
                'maxResults': MAX_NUM_IMAGE_DETECTION
            }]
        }]
    })
    response = service_request.execute()
    
    detectFlag = False
    detectList = []
    annotations = None

    #
    size = len(response['responses'][0])
    print "size:" + str(size)
    if size >= 1:
        #        
        if mode == 'LABEL_DETECTION':
            annotations = response['responses'][0]['labelAnnotations']
            for label_num in range(0, len(annotations)):
                label = annotations[label_num]['description']
                score = annotations[label_num]['score']
                print('Found label: %s, score: %f' % (label, score))
                detectList.append(label)
                if compStr in label:
                    detectFlag = True
                    reStr = compStr
                    break
            if detectFlag == False:
                reStr = detectList[0]
                
            return reStr
        #
        elif mode == 'LOGO_DETECTION':
            annotations = response['responses'][0]['logoAnnotations']
            for label_num in range(0, len(annotations)):
                logo = annotations[label_num]['description']
                score = annotations[label_num]['score']
                print('Found logo: %s, score: %f' % (logo, score))
                detectList.append(logo)
                if compStr in logo:
                    detectFlag = True
                    reStr = compStr
                    break
            if detectFlag == False:
                reStr = detectList[0]
        #
        elif mode == 'FACE_DETECTION':
            annotations = response['responses'][0]['faceAnnotations']
            joyVal = annotations[0]['joyLikelihood']
            angVal = annotations[0]['angerLikelihood']
            sorVal = annotations[0]['sorrowLikelihood']
            sprVal = annotations[0]['surpriseLikelihood']
            print ('joyVal %s' % joyVal)
            print ('angVal %s' % angVal)
            print ('sorVal %s' % sorVal)
            print ('sprVal %s' % sprVal)
            if joyVal == 'VERY_LIKELY' or joyVal == 'LIKELY' or joyVal == 'POSSIBLE':
                reStr = 'JOY'
            elif angVal == 'VERY_LIKELY' or angVal == 'LIKELY' or joyVal == 'POSSIBLE':
                reStr = 'ANGER'
            elif sorVal == 'VERY_LIKELY' or sorVal == 'LIKELY' or joyVal == 'POSSIBLE':
                reStr = 'SORROW'
            elif sprVal == 'VERY_LIKELY' or sprVal == 'LIKELY' or joyVal == 'POSSIBLE':
                reStr = 'SURPRISE'
            else :
                reStr = 'NORMAL'
        #
        elif mode == 'TEXT_DETECTION':
            annotations = response['responses'][0]['textAnnotations']
            txt = annotations[0]['description']
            print ('txt:%s' % txt )
            if compStr in txt:
                reStr = compStr
            else: 
                reStr = '-1'
       
        #
        print annotations[0]['boundingPoly']['vertices']
        if annotations[0]['boundingPoly']['vertices'][0].has_key('x'):
            rect[0] = annotations[0]['boundingPoly']['vertices'][0]['x']
        else :
            rect[0] = 0
        if annotations[0]['boundingPoly']['vertices'][0].has_key('y'):
            rect[1] = annotations[0]['boundingPoly']['vertices'][0]['y']
        else :
            rect[1] = 0
        rect[2] = annotations[0]['boundingPoly']['vertices'][2]['x']
        rect[3] = annotations[0]['boundingPoly']['vertices'][2]['y']

        #
        cv2.rectangle(img, (rect[0],rect[1]), (rect[2],rect[3]), (255,255,255), 5)
        cv2.imshow('nametag1.jpg', img)
        cv2.waitKey(0)
        cv2.destroyAllWindows()

        return reStr
    #
    else :
        return '-1'

        
if __name__ == '__main__':

    argvs = sys.argv  # コマンドライン引数を格納したリストの取得
    argc = len(argvs) # 引数の個数
    if argc == 2:
        inputFlag = True
        filename = argvs[1]
        
    rect = []

    print 'FACE_DETECTION'
    text = imageDetermin('FACE_DETECTION', 'JOY', rect)
    print text
    print rect
    print ''

    print 'LABEL_DETECTION'
    text = imageDetermin('LABEL_DETECTION', 'dog', rect)
    print text
    print rect
    print ''

    print 'TEXT_DETECTION'
    text = imageDetermin('TEXT_DETECTION', 'Staff', rect)
    print text
    print rect
    print ''
 
    print 'LOGO_DETECTION'
    text = imageDetermin('LOGO_DETECTION', 'panasonic', rect)
    print text
    print rect

