Speed up predictions for Object Detection

Question

I am struggling to get good FPS for my predictions. I am running my predictions on a Tesla K80 and I'd like to speed up my predictions by at least a factor of 20. Here is my code:

def load_detection_graph(PATH_TO_CKPT):
       detection_graph = tf.Graph()
        with detection_graph.as_default():
          od_graph_def = tf.GraphDef()
          with tf.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='')
        return detection_graph



    def load_image_into_numpy_array(image):
        '''
        convert image to numpy arrays
        '''
        (im_width, im_height) = image.size
        return np.array(image.getdata()).reshape((im_height, im_width, 3)).astype(np.uint8)



    def run_inference_for_single_image(image, graph, filename):
      with graph.as_default():
        with tf.Session() as sess:
          # Get handles to input and output tensors
          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)
          if 'detection_masks' in tensor_dict:
            # The following processing is only for single image
            detection_boxes = tf.squeeze(tensor_dict['detection_boxes'], [0])
            detection_masks = tf.squeeze(tensor_dict['detection_masks'], [0])
            # Reframe is required to translate mask from box coordinates to image coordinates and fit the image size.
            real_num_detection = tf.cast(tensor_dict['num_detections'][0], tf.int32)
            detection_boxes = tf.slice(detection_boxes, [0, 0], [real_num_detection, -1])
            detection_masks = tf.slice(detection_masks, [0, 0, 0], [real_num_detection, -1, -1])
            detection_masks_reframed = utils_ops.reframe_box_masks_to_image_masks(
                detection_masks, detection_boxes, image.shape[0], image.shape[1])
            detection_masks_reframed = tf.cast(
                tf.greater(detection_masks_reframed, 0.5), tf.uint8)
            # Follow the convention by adding back the batch dimension
            tensor_dict['detection_masks'] = tf.expand_dims(
                detection_masks_reframed, 0)
          image_tensor = tf.get_default_graph().get_tensor_by_name('image_tensor:0')

          # Run inference
          output_dict = sess.run(tensor_dict,
                                 feed_dict={image_tensor: np.expand_dims(image, 0)})

          # all outputs are float32 numpy arrays, so convert types as appropriate
          output_dict['filename'] = filename
          output_dict['num_detections'] = int(output_dict['num_detections'][0])
          output_dict['detection_classes'] = output_dict[
              'detection_classes'][0].astype(np.uint8)
          output_dict['detection_boxes'] = output_dict['detection_boxes'][0]
          output_dict['detection_scores'] = output_dict['detection_scores'][0]
          if 'detection_masks' in output_dict:
            output_dict['detection_masks'] = output_dict['detection_masks'][0]
      return output_dict


    def predict_image(TEST_IMAGE_PATHS, PATH_TO_CKPT, category_index, save_path):
        detection_graph = load_detection_graph(PATH_TO_CKPT)
        prediction_dict = defaultdict()
        start_time = time.time()
        for image_path in TEST_IMAGE_PATHS:
            toc = time.time()
            filename = image_path
            image = Image.open(image_path)
            # the array based representation of the image will be used later in order to prepare the
            # result image with boxes and labels on it.
            image_np = load_image_into_numpy_array(image)
            # Expand dimensions since the model expects images to have shape: [1, None, None, 3]
            image_np_expanded = np.expand_dims(image_np, axis=0)
            # Actual detection.
            output_dict = run_inference_for_single_image(image_np, detection_graph, filename)
            # Visualization of the results of a detection.

            vis_util.visualize_boxes_and_labels_on_image_array(
              image_np,
              output_dict['detection_boxes'],
              output_dict['detection_classes'],
              output_dict['detection_scores'],
              category_index,
              instance_masks=output_dict.get('detection_masks'),
              use_normalized_coordinates=True,
              line_thickness=1)
            prediction_dict[filename] = output_dict
            plt.figure(figsize=(8,6), dpi=100)
            plt.imshow(image_np)
            plt.savefig(save_path+'{}'.format(filename))
            tic = time.time()
            print('{0} saved in {1:.2f}sec'.format(filename, tic-toc))
        end_time = time.time()
        print('{0:.2f}min to predict all images'.format((end_time-start_time)/60))
        with open('../predictions/predictions.pickle', 'wb') as f:
            pickle.dump(prediction_dict, f)
        return prediction_dict

Right now I am getting about 1.8 sec per detection. That includes saving image and drawing bounding boxes. I do not need to save image or draw bounding boxes, I just need the output_dict. Any advice on how to speed this up?

Srinivas Bringu · Accepted Answer

Session creation is the most costly operation, dont re-create it everytime, try to re-use the session object

Speed up predictions for Object Detection

Answers (2)

Related Questions