def worker(input_q, output_q, cap_params, frame_processed):
print(">> Loading frozen model for worker.")
detection_graph, sess = detector_utils.load_inference_graph()
sess = tf.compat.v1.Session(graph=detection_graph)
while True:
frame = input_q.get()
if (frame is not None):
'''Boxes contain coordinates for detected hands
Scores contains condfidence levels
If len(boxes) > 1, at least one hand is detected
You can change the score_thresh value as desired'''
boxes, scores = detector_utils.detect_objects(
frame, detection_graph, sess)
# Draws bounding boxes
detector_utils.draw_box_on_image(cap_params['num_hands_detect'],
cap_params["score_thresh"],
scores, boxes,
cap_params['im_width'],
cap_params['im_height'], frame)
# Adds frame annotated with bounding box to queue
output_q.put(frame)
frame_processed += 1
else:
output_q.put(frame)
sess.close()
def worker_hands(input_q, output_q):
detection_graph, sess = detector_utils.load_inference_graph()
sess = tf.Session(graph=detection_graph)
while True:
frame = input_q.get()
if frame is not None:
boxes, scores = detector_utils.detect_objects(
frame, detection_graph, sess)
output_q.put((boxes, scores))
else:
output_q.put((boxes, scores))
sess.close()
box_padding = 0.12 checkpoint_path = '../extra/model/east_icdar2015_resnet_v1_50_rbox' #get the start time start_time = datetime.datetime.now() #max number of hands we want to detect num_hands_detect = 1 score_thresh = 0.4 timer_start = 0 #load hands detection model detection_graph, sess1 = detector_utils.load_inference_graph() timer = 0 #flag indicating whether the patient finish the whole process finished = 0 hands = [] def main(argv=None): global im process_start = time.time() #location of the instructions
import numpy as np
import keras
from keras.models import Model, load_model
from keras.preprocessing import image as KerasImage
import cv2
import string
import random
import requests
import os
TrashRoutes = Blueprint('TrashRoutes', __name__)
detection_graph, session = load_inference_graph()
MIN_THRESHOLD = 0.5
prediction_list=['cardboard', 'glass', 'trash', 'paper', 'plastic', 'trash']
# with CustomObjectScope({'relu6': ReLU,'DepthwiseConv2D': DepthwiseConv2D}):
model=load_model('trained_model.h5')
model._make_predict_function()
labels={0: 'cardboard', 1: 'glass', 2: 'trash', 3: 'paper', 4: 'plastic', 5: 'trash'}
@TrashRoutes.route('/trashAll/', methods=["GET"])
def get_all_trash():
trash = Trash.query.all()
return jsonify([t.serialize() for t in trash])
def __init__(self):
self.detection_graph, self.sess = detector_utils.load_inference_graph()