Object Detection Tensorflow

import matplotlib
matplotlib.use('TkAgg')

import ntpath
import numpy as np
import os
import six.moves.urllib as urllib
import sys
import tarfile
import tensorflow as tf
import zipfile
import cv2

from distutils.version import StrictVersion
from collections import defaultdict
from io import StringIO
from matplotlib import pyplot as plt
from PIL import Image
from glob import glob


sys.path.append("/media/quazi/DATADRIVE1/tensorflow/models/research/")
sys.path.append("/media/quazi/DATADRIVE1/tensorflow/models/research/object_detection/")
from object_detection.utils import ops as utils_ops

if StrictVersion(tf.__version__) < StrictVersion('1.9.0'):
	raise ImportError('Please upgrade your TensorFlow installation to v1.9.* or later!')


from utils import label_map_util
from utils import visualization_utils as vis_util



class ObjectDetection:
	def __init__(self, path_to_frozen_graph = None, path_to_labels = None):
		self.path_to_frozen_graph = path_to_frozen_graph #'/media/quazi/DATADRIVE1/data/tf_models/faster_rcnn_resnet101_kitti_2018_01_28/frozen_inference_graph.pb'
		self.path_to_labels = path_to_labels  #= '/media/quazi/DATADRIVE1/data/tf_models/faster_rcnn_resnet101_kitti_2018_01_28/kitti_label_map.pbtxt'


		self.category_index = label_map_util.create_category_index_from_labelmap(self.path_to_labels, use_display_name=True)
		self.detection_graph = tf.Graph()
		with self.detection_graph.as_default():
			od_graph_def = tf.GraphDef()
			with tf.gfile.GFile(self.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 read_images(self, paths, resize=True):
		images = []
		for path in paths:
			img = cv2.imread(path)
			# img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
			if resize:
				img = cv2.resize(img, (300, 300))
			images.append(img)
		images = np.array(images)
		return images

	def run_inference_for_multiple_images(self, image_paths, batch_size=16):
		
		detections = list()
		
		with self.detection_graph.as_default():
			with tf.Session() as sess:
				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)


				for i in range(0, len(image_paths), batch_size):
					image_paths_batch = image_paths[i:i+batch_size]
					images = self.read_images(image_paths_batch)

					image_tensor = tf.get_default_graph().get_tensor_by_name('image_tensor:0')	
					output_dict = sess.run(tensor_dict, feed_dict={image_tensor: images})
					
					print (images.shape)
					for j in range(images.shape[0]):
						dd = dict()
						dd['num_detections'] = int(output_dict['num_detections'][j])
						dd['detection_classes'] = output_dict['detection_classes'][j].astype(np.int64)
						dd['detection_boxes'] = output_dict['detection_boxes'][j]
						dd['detection_scores'] = output_dict['detection_scores'][j]
						detections.append(dd)

		return detections



object_detection = ObjectDetection(
		path_to_frozen_graph = '/media/quazi/DATADRIVE1/data/ssd_inception_v2_coco/ssd_inception_v2_coco_2018_01_28/frozen_inference_graph.pb',
		path_to_labels = '/media/quazi/DATADRIVE1/data/ssd_inception_v2_coco/mscoco_label_map.pbtxt')

image_paths = glob('/media/quazi/DATADRIVE1/data/coco/val2017/*.jpg')[:10]
detections = object_detection.run_inference_for_multiple_images(image_paths, batch_size=16)


for id, (image_path, detection) in enumerate(zip(image_paths, detections)):
	img = object_detection.read_images([image_path], resize=False)[0]
	img_height, img_width, _ = img.shape
	for obj_class, obj_bbox, obj_score in zip(detection['detection_classes'], detection['detection_boxes'], detection['detection_scores']):
		if obj_score >= 0.5:
			ymin, xmin, ymax, xmax = obj_bbox
			
			ymin = int(ymin * img_height)
			ymax = int(ymax * img_height)
			xmin = int(xmin * img_width)
			xmax = int(xmax * img_width)
			
			img = cv2.rectangle(img, (xmin, ymin), (xmax, ymax), (255, 0, 0), 5)

	cv2.imwrite('./img_%d.jpg'%id, img)