from imutils.video import VideoStream
from imutils.video import FPS
from imutils.video import FileVideoStream
import numpy as np
import argparse
import imutils
import time
import cv2
import os
import json
import sys
from openvino.inference_engine import IENetwork
from openvino.inference_engine import IECore

print("load model in NCS....")
ie = IECore()
net = ie.read_network(model="person-detection-action-recognition-0006.xml", weights="person-detection-action-recognition-0006.bin")
print (net.outputs)
input_blob = next(iter(net.inputs))
out_blob_h1a1 = "ActionNet/action_heads/out_head_1_anchor_1"
out_blob_h2a1 = "ActionNet/action_heads/out_head_2_anchor_1"
out_blob_h2a2 = "ActionNet/action_heads/out_head_2_anchor_2"
out_blob_h2a3 = "ActionNet/action_heads/out_head_2_anchor_3"
out_blob_h2a4 = "ActionNet/action_heads/out_head_2_anchor_4"
out_blob_conf = "ActionNet/out_detection_conf"
out_blob_loc = "ActionNet/out_detection_loc"


#  Load network to the plugin
exec_net = ie.load_network(network=net, device_name="MYRIAD")


print("model loaded correctly...")

print("[INFO] starting video stream...")
# picamera code
# file code
# cap = cv2.VideoCapture('/home/pi/Desktop/cogito-person-vehicle-recognition/test.mp4')

# vs = FileVideoStream("/home/pi/Desktop/cogito-person-vehicle-recognition/test.m4v")
# cap = cv2.VideoCapture('/home/pi/Desktop/cogito-person-vehicle-recognition/test_Cropped_last.m4v')
time.sleep(2.0)
#fps = FPS().start()
initial_w = 600
initial_h = 450
# f = open("timewithmovidius.txt", "a")
cont = 0



frame = cv2.imread("test.jpg")
# ret, frame = cap.read()
# comment this if the video is in streaming
# cont = cont+1
# if cont % 15 !=0:
# continue
# end comment

# ret, frame = cap.read()
#frame = imutils.resize(frame, width=600, height=450)
# create input for the neural network
n, c, h, w = net.inputs[input_blob].shape
image = cv2.resize(frame, (w, h))
image = image.transpose((2, 0, 1))
# Change data layout from HWC to CHW
image = image.reshape((n, c, h, w))
# start_time = time.clock()
req_handle = exec_net.start_async(request_id=0, inputs={input_blob: image})
status = req_handle.wait()
res = req_handle.outputs[out_blob_h1a1]
#np.savetxt("h1a1.txt",res[0])
file_1 = open("h1a1.txt","w")
print ("shape h1a1")
print (res,file=file_1)
file_1.close()



res = req_handle.outputs[out_blob_h2a1]
#np.savetxt("h1a1.txt",res[0])
file_1 = open("h2a1.txt","w")
print ("shape h1a1")
print (res,file=file_1)
file_1.close()

res = req_handle.outputs[out_blob_h2a2]
#np.savetxt("h1a1.txt",res[0])
file_1 = open("h2a2.txt","w")
print (res,file=file_1)
file_1.close()

res = req_handle.outputs[out_blob_h2a3]
#np.savetxt("h1a1.txt",res[0])
file_1 = open("h2a3.txt","w")
print (res,file=file_1)
file_1.close()

res = req_handle.outputs[out_blob_h2a4]
#np.savetxt("h1a1.txt",res[0])
file_1 = open("h2a4.txt","w")
print (res,file=file_1)
file_1.close()

res = req_handle.outputs[out_blob_loc]
#np.savetxt("h1a1.txt",res[0])
file_1 = open("location.txt","w")
print (res,file=file_1)
file_1.close()

res = req_handle.outputs[out_blob_conf]
#np.savetxt("h1a1.txt",res[0])
file_1 = open("conf.txt","w")
print (res,file=file_1)
file_1.close()





exit()