eval_util.py

# Copyright 2017 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
"""Common utility functions for evaluation."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import collections
import os
import re
import time

import numpy as np
from six.moves import range
import tensorflow.compat.v1 as tf

import tf_slim as slim

from object_detection.core import box_list
from object_detection.core import box_list_ops
from object_detection.core import keypoint_ops
from object_detection.core import standard_fields as fields
from object_detection.metrics import coco_evaluation
from object_detection.protos import eval_pb2
from object_detection.utils import label_map_util
from object_detection.utils import object_detection_evaluation
from object_detection.utils import ops
from object_detection.utils import shape_utils
from object_detection.utils import visualization_utils as vis_utils

EVAL_KEYPOINT_METRIC = 'coco_keypoint_metrics'

# A dictionary of metric names to classes that implement the metric. The classes
# in the dictionary must implement
# utils.object_detection_evaluation.DetectionEvaluator interface.
EVAL_METRICS_CLASS_DICT = {
    'coco_detection_metrics':
        coco_evaluation.CocoDetectionEvaluator,
    'coco_keypoint_metrics':
        coco_evaluation.CocoKeypointEvaluator,
    'coco_mask_metrics':
        coco_evaluation.CocoMaskEvaluator,
    'oid_challenge_detection_metrics':
        object_detection_evaluation.OpenImagesDetectionChallengeEvaluator,
    'oid_challenge_segmentation_metrics':
        object_detection_evaluation
        .OpenImagesInstanceSegmentationChallengeEvaluator,
    'pascal_voc_detection_metrics':
        object_detection_evaluation.PascalDetectionEvaluator,
    'weighted_pascal_voc_detection_metrics':
        object_detection_evaluation.WeightedPascalDetectionEvaluator,
    'precision_at_recall_detection_metrics':
        object_detection_evaluation.PrecisionAtRecallDetectionEvaluator,
    'pascal_voc_instance_segmentation_metrics':
        object_detection_evaluation.PascalInstanceSegmentationEvaluator,
    'weighted_pascal_voc_instance_segmentation_metrics':
        object_detection_evaluation.WeightedPascalInstanceSegmentationEvaluator,
    'oid_V2_detection_metrics':
        object_detection_evaluation.OpenImagesDetectionEvaluator,
}

EVAL_DEFAULT_METRIC = 'coco_detection_metrics'


def write_metrics(metrics, global_step, summary_dir):
  """Write metrics to a summary directory.

  Args:
    metrics: A dictionary containing metric names and values.
    global_step: Global step at which the metrics are computed.
    summary_dir: Directory to write tensorflow summaries to.
  """
  tf.logging.info('Writing metrics to tf summary.')
  summary_writer = tf.summary.FileWriterCache.get(summary_dir)
  for key in sorted(metrics):
    summary = tf.Summary(value=[
        tf.Summary.Value(tag=key, simple_value=metrics[key]),
    ])
    summary_writer.add_summary(summary, global_step)
    tf.logging.info('%s: %f', key, metrics[key])
  tf.logging.info('Metrics written to tf summary.')


# TODO(rathodv): Add tests.
def visualize_detection_results(result_dict,
                                tag,
                                global_step,
                                categories,
                                summary_dir='',
                                export_dir='',
                                agnostic_mode=False,
                                show_groundtruth=False,
                                groundtruth_box_visualization_color='black',
                                min_score_thresh=.5,
                                max_num_predictions=20,
                                skip_scores=False,
                                skip_labels=False,
                                keep_image_id_for_visualization_export=False):
  """Visualizes detection results and writes visualizations to image summaries.

  This function visualizes an image with its detected bounding boxes and writes
  to image summaries which can be viewed on tensorboard.  It optionally also
  writes images to a directory. In the case of missing entry in the label map,
  unknown class name in the visualization is shown as "N/A".

  Args:
    result_dict: a dictionary holding groundtruth and detection
      data corresponding to each image being evaluated.  The following keys
      are required:
        'original_image': a numpy array representing the image with shape
          [1, height, width, 3] or [1, height, width, 1]
        'detection_boxes': a numpy array of shape [N, 4]
        'detection_scores': a numpy array of shape [N]
        'detection_classes': a numpy array of shape [N]
      The following keys are optional:
        'groundtruth_boxes': a numpy array of shape [N, 4]
        'groundtruth_keypoints': a numpy array of shape [N, num_keypoints, 2]
      Detections are assumed to be provided in decreasing order of score and for
      display, and we assume that scores are probabilities between 0 and 1.
    tag: tensorboard tag (string) to associate with image.
    global_step: global step at which the visualization are generated.
    categories: a list of dictionaries representing all possible categories.
      Each dict in this list has the following keys:
          'id': (required) an integer id uniquely identifying this category
          'name': (required) string representing category name
            e.g., 'cat', 'dog', 'pizza'
          'supercategory': (optional) string representing the supercategory
            e.g., 'animal', 'vehicle', 'food', etc
    summary_dir: the output directory to which the image summaries are written.
    export_dir: the output directory to which images are written.  If this is
      empty (default), then images are not exported.
    agnostic_mode: boolean (default: False) controlling whether to evaluate in
      class-agnostic mode or not.
    show_groundtruth: boolean (default: False) controlling whether to show
      groundtruth boxes in addition to detected boxes
    groundtruth_box_visualization_color: box color for visualizing groundtruth
      boxes
    min_score_thresh: minimum score threshold for a box to be visualized
    max_num_predictions: maximum number of detections to visualize
    skip_scores: whether to skip score when drawing a single detection
    skip_labels: whether to skip label when drawing a single detection
    keep_image_id_for_visualization_export: whether to keep image identifier in
      filename when exported to export_dir
  Raises:
    ValueError: if result_dict does not contain the expected keys (i.e.,
      'original_image', 'detection_boxes', 'detection_scores',
      'detection_classes')
  """
  detection_fields = fields.DetectionResultFields
  input_fields = fields.InputDataFields
  if not set([
      input_fields.original_image,
      detection_fields.detection_boxes,
      detection_fields.detection_scores,
      detection_fields.detection_classes,
  ]).issubset(set(result_dict.keys())):
    raise ValueError('result_dict does not contain all expected keys.')
  if show_groundtruth and input_fields.groundtruth_boxes not in result_dict:
    raise ValueError('If show_groundtruth is enabled, result_dict must contain '
                     'groundtruth_boxes.')
  tf.logging.info('Creating detection visualizations.')
  category_index = label_map_util.create_category_index(categories)

  image = np.squeeze(result_dict[input_fields.original_image], axis=0)
  if image.shape[2] == 1:  # If one channel image, repeat in RGB.
    image = np.tile(image, [1, 1, 3])
  detection_boxes = result_dict[detection_fields.detection_boxes]
  detection_scores = result_dict[detection_fields.detection_scores]
  detection_classes = np.int32((result_dict[
      detection_fields.detection_classes]))
  detection_keypoints = result_dict.get(detection_fields.detection_keypoints)
  detection_masks = result_dict.get(detection_fields.detection_masks)
  detection_boundaries = result_dict.get(detection_fields.detection_boundaries)

  # Plot groundtruth underneath detections
  if show_groundtruth:
    groundtruth_boxes = result_dict[input_fields.groundtruth_boxes]
    groundtruth_keypoints = result_dict.get(input_fields.groundtruth_keypoints)
    vis_utils.visualize_boxes_and_labels_on_image_array(
        image=image,
        boxes=groundtruth_boxes,
        classes=None,
        scores=None,
        category_index=category_index,
        keypoints=groundtruth_keypoints,
        use_normalized_coordinates=False,
        max_boxes_to_draw=None,
        groundtruth_box_visualization_color=groundtruth_box_visualization_color)
  vis_utils.visualize_boxes_and_labels_on_image_array(
      image,
      detection_boxes,
      detection_classes,
      detection_scores,
      category_index,
      instance_masks=detection_masks,
      instance_boundaries=detection_boundaries,
      keypoints=detection_keypoints,
      use_normalized_coordinates=False,
      max_boxes_to_draw=max_num_predictions,
      min_score_thresh=min_score_thresh,
      agnostic_mode=agnostic_mode,
      skip_scores=skip_scores,
      skip_labels=skip_labels)

  if export_dir:
    if keep_image_id_for_visualization_export and result_dict[fields.
                                                              InputDataFields()
                                                              .key]:
      export_path = os.path.join(export_dir, 'export-{}-{}.png'.format(
          tag, result_dict[fields.InputDataFields().key]))
    else:
      export_path = os.path.join(export_dir, 'export-{}.png'.format(tag))
    vis_utils.save_image_array_as_png(image, export_path)

  summary = tf.Summary(value=[
      tf.Summary.Value(
          tag=tag,
          image=tf.Summary.Image(
              encoded_image_string=vis_utils.encode_image_array_as_png_str(
                  image)))
  ])
  summary_writer = tf.summary.FileWriterCache.get(summary_dir)
  summary_writer.add_summary(summary, global_step)

  tf.logging.info('Detection visualizations written to summary with tag %s.',
                  tag)


def _run_checkpoint_once(tensor_dict,
                         evaluators=None,
                         batch_processor=None,
                         checkpoint_dirs=None,
                         variables_to_restore=None,
                         restore_fn=None,
                         num_batches=1,
                         master='',
                         save_graph=False,
                         save_graph_dir='',
                         losses_dict=None,
                         eval_export_path=None,
                         process_metrics_fn=None):
  """Evaluates metrics defined in evaluators and returns summaries.

  This function loads the latest checkpoint in checkpoint_dirs and evaluates
  all metrics defined in evaluators. The metrics are processed in batch by the
  batch_processor.

  Args:
    tensor_dict: a dictionary holding tensors representing a batch of detections
      and corresponding groundtruth annotations.
    evaluators: a list of object of type DetectionEvaluator to be used for
      evaluation. Note that the metric names produced by different evaluators
      must be unique.
    batch_processor: a function taking four arguments:
      1. tensor_dict: the same tensor_dict that is passed in as the first
        argument to this function.
      2. sess: a tensorflow session
      3. batch_index: an integer representing the index of the batch amongst
        all batches
      By default, batch_processor is None, which defaults to running:
        return sess.run(tensor_dict)
      To skip an image, it suffices to return an empty dictionary in place of
      result_dict.
    checkpoint_dirs: list of directories to load into an EnsembleModel. If it
      has only one directory, EnsembleModel will not be used --
        a DetectionModel
      will be instantiated directly. Not used if restore_fn is set.
    variables_to_restore: None, or a dictionary mapping variable names found in
      a checkpoint to model variables. The dictionary would normally be
      generated by creating a tf.train.ExponentialMovingAverage object and
      calling its variables_to_restore() method. Not used if restore_fn is set.
    restore_fn: None, or a function that takes a tf.Session object and correctly
      restores all necessary variables from the correct checkpoint file. If
      None, attempts to restore from the first directory in checkpoint_dirs.
    num_batches: the number of batches to use for evaluation.
    master: the location of the Tensorflow session.
    save_graph: whether or not the Tensorflow graph is stored as a pbtxt file.
    save_graph_dir: where to store the Tensorflow graph on disk. If save_graph
      is True this must be non-empty.
    losses_dict: optional dictionary of scalar detection losses.
    eval_export_path: Path for saving a json file that contains the detection
      results in json format.
    process_metrics_fn: a callback called with evaluation results after each
      evaluation is done.  It could be used e.g. to back up checkpoints with
      best evaluation scores, or to call an external system to update evaluation
      results in order to drive best hyper-parameter search.  Parameters are:
      int checkpoint_number, Dict[str, ObjectDetectionEvalMetrics] metrics,
      str checkpoint_file path.

  Returns:
    global_step: the count of global steps.
    all_evaluator_metrics: A dictionary containing metric names and values.

  Raises:
    ValueError: if restore_fn is None and checkpoint_dirs doesn't have at least
      one element.
    ValueError: if save_graph is True and save_graph_dir is not defined.
  """
  if save_graph and not save_graph_dir:
    raise ValueError('`save_graph_dir` must be defined.')
  sess = tf.Session(master, graph=tf.get_default_graph())
  sess.run(tf.global_variables_initializer())
  sess.run(tf.local_variables_initializer())
  sess.run(tf.tables_initializer())
  checkpoint_file = None
  if restore_fn:
    restore_fn(sess)
  else:
    if not checkpoint_dirs:
      raise ValueError('`checkpoint_dirs` must have at least one entry.')
    checkpoint_file = tf.train.latest_checkpoint(checkpoint_dirs[0])
    saver = tf.train.Saver(variables_to_restore)
    saver.restore(sess, checkpoint_file)

  if save_graph:
    tf.train.write_graph(sess.graph_def, save_graph_dir, 'eval.pbtxt')

  counters = {'skipped': 0, 'success': 0}
  aggregate_result_losses_dict = collections.defaultdict(list)
  with slim.queues.QueueRunners(sess):
    try:
      for batch in range(int(num_batches)):
        if (batch + 1) % 100 == 0:
          tf.logging.info('Running eval ops batch %d/%d', batch + 1,
                          num_batches)
        if not batch_processor:
          try:
            if not losses_dict:
              losses_dict = {}
            result_dict, result_losses_dict = sess.run([tensor_dict,
                                                        losses_dict])
            counters['success'] += 1
          except tf.errors.InvalidArgumentError:
            tf.logging.info('Skipping image')
            counters['skipped'] += 1
            result_dict = {}
        else:
          result_dict, result_losses_dict = batch_processor(
              tensor_dict, sess, batch, counters, losses_dict=losses_dict)
        if not result_dict:
          continue
        for key, value in iter(result_losses_dict.items()):
          aggregate_result_losses_dict[key].append(value)
        for evaluator in evaluators:
          # TODO(b/65130867): Use image_id tensor once we fix the input data
          # decoders to return correct image_id.
          # TODO(akuznetsa): result_dict contains batches of images, while
          # add_single_ground_truth_image_info expects a single image. Fix
          if (isinstance(result_dict, dict) and
              fields.InputDataFields.key in result_dict and
              result_dict[fields.InputDataFields.key]):
            image_id = result_dict[fields.InputDataFields.key]
          else:
            image_id = batch
          evaluator.add_single_ground_truth_image_info(
              image_id=image_id, groundtruth_dict=result_dict)
          evaluator.add_single_detected_image_info(
              image_id=image_id, detections_dict=result_dict)
      tf.logging.info('Running eval batches done.')
    except tf.errors.OutOfRangeError:
      tf.logging.info('Done evaluating -- epoch limit reached')
    finally:
      # When done, ask the threads to stop.
      tf.logging.info('# success: %d', counters['success'])
      tf.logging.info('# skipped: %d', counters['skipped'])
      all_evaluator_metrics = {}
      if eval_export_path and eval_export_path is not None:
        for evaluator in evaluators:
          if (isinstance(evaluator, coco_evaluation.CocoDetectionEvaluator) or
              isinstance(evaluator, coco_evaluation.CocoMaskEvaluator)):
            tf.logging.info('Started dumping to json file.')
            evaluator.dump_detections_to_json_file(
                json_output_path=eval_export_path)
            tf.logging.info('Finished dumping to json file.')
      for evaluator in evaluators:
        metrics = evaluator.evaluate()
        evaluator.clear()
        if any(key in all_evaluator_metrics for key in metrics):
          raise ValueError('Metric names between evaluators must not collide.')
        all_evaluator_metrics.update(metrics)
      global_step = tf.train.global_step(sess, tf.train.get_global_step())

      for key, value in iter(aggregate_result_losses_dict.items()):
        all_evaluator_metrics['Losses/' + key] = np.mean(value)
      if process_metrics_fn and checkpoint_file:
        m = re.search(r'model.ckpt-(\d+)$', checkpoint_file)
        if not m:
          tf.logging.error('Failed to parse checkpoint number from: %s',
                           checkpoint_file)
        else:
          checkpoint_number = int(m.group(1))
          process_metrics_fn(checkpoint_number, all_evaluator_metrics,
                             checkpoint_file)
  sess.close()
  return (global_step, all_evaluator_metrics)


# TODO(rathodv): Add tests.
def repeated_checkpoint_run(tensor_dict,
                            summary_dir,
                            evaluators,
                            batch_processor=None,
                            checkpoint_dirs=None,
                            variables_to_restore=None,
                            restore_fn=None,
                            num_batches=1,
                            eval_interval_secs=120,
                            max_number_of_evaluations=None,
                            max_evaluation_global_step=None,
                            master='',
                            save_graph=False,
                            save_graph_dir='',
                            losses_dict=None,
                            eval_export_path=None,
                            process_metrics_fn=None):
  """Periodically evaluates desired tensors using checkpoint_dirs or restore_fn.

  This function repeatedly loads a checkpoint and evaluates a desired
  set of tensors (provided by tensor_dict) and hands the resulting numpy
  arrays to a function result_processor which can be used to further
  process/save/visualize the results.

  Args:
    tensor_dict: a dictionary holding tensors representing a batch of detections
      and corresponding groundtruth annotations.
    summary_dir: a directory to write metrics summaries.
    evaluators: a list of object of type DetectionEvaluator to be used for
      evaluation. Note that the metric names produced by different evaluators
      must be unique.
    batch_processor: a function taking three arguments:
      1. tensor_dict: the same tensor_dict that is passed in as the first
        argument to this function.
      2. sess: a tensorflow session
      3. batch_index: an integer representing the index of the batch amongst
        all batches
      By default, batch_processor is None, which defaults to running:
        return sess.run(tensor_dict)
    checkpoint_dirs: list of directories to load into a DetectionModel or an
      EnsembleModel if restore_fn isn't set. Also used to determine when to run
      next evaluation. Must have at least one element.
    variables_to_restore: None, or a dictionary mapping variable names found in
      a checkpoint to model variables. The dictionary would normally be
      generated by creating a tf.train.ExponentialMovingAverage object and
      calling its variables_to_restore() method. Not used if restore_fn is set.
    restore_fn: a function that takes a tf.Session object and correctly restores
      all necessary variables from the correct checkpoint file.
    num_batches: the number of batches to use for evaluation.
    eval_interval_secs: the number of seconds between each evaluation run.
    max_number_of_evaluations: the max number of iterations of the evaluation.
      If the value is left as None the evaluation continues indefinitely.
    max_evaluation_global_step: global step when evaluation stops.
    master: the location of the Tensorflow session.
    save_graph: whether or not the Tensorflow graph is saved as a pbtxt file.
    save_graph_dir: where to save on disk the Tensorflow graph. If store_graph
      is True this must be non-empty.
    losses_dict: optional dictionary of scalar detection losses.
    eval_export_path: Path for saving a json file that contains the detection
      results in json format.
    process_metrics_fn: a callback called with evaluation results after each
      evaluation is done.  It could be used e.g. to back up checkpoints with
      best evaluation scores, or to call an external system to update evaluation
      results in order to drive best hyper-parameter search.  Parameters are:
      int checkpoint_number, Dict[str, ObjectDetectionEvalMetrics] metrics,
      str checkpoint_file path.

  Returns:
    metrics: A dictionary containing metric names and values in the latest
      evaluation.

  Raises:
    ValueError: if max_num_of_evaluations is not None or a positive number.
    ValueError: if checkpoint_dirs doesn't have at least one element.
  """
  if max_number_of_evaluations and max_number_of_evaluations <= 0:
    raise ValueError(
        '`max_number_of_evaluations` must be either None or a positive number.')
  if max_evaluation_global_step and max_evaluation_global_step <= 0:
    raise ValueError(
        '`max_evaluation_global_step` must be either None or positive.')

  if not checkpoint_dirs:
    raise ValueError('`checkpoint_dirs` must have at least one entry.')

  last_evaluated_model_path = None
  number_of_evaluations = 0
  while True:
    start = time.time()
    tf.logging.info('Starting evaluation at ' + time.strftime(
        '%Y-%m-%d-%H:%M:%S', time.gmtime()))
    model_path = tf.train.latest_checkpoint(checkpoint_dirs[0])
    if not model_path:
      tf.logging.info('No model found in %s. Will try again in %d seconds',
                      checkpoint_dirs[0], eval_interval_secs)
    elif model_path == last_evaluated_model_path:
      tf.logging.info('Found already evaluated checkpoint. Will try again in '
                      '%d seconds', eval_interval_secs)
    else:
      last_evaluated_model_path = model_path
      global_step, metrics = _run_checkpoint_once(
          tensor_dict,
          evaluators,
          batch_processor,
          checkpoint_dirs,
          variables_to_restore,
          restore_fn,
          num_batches,
          master,
          save_graph,
          save_graph_dir,
          losses_dict=losses_dict,
          eval_export_path=eval_export_path,
          process_metrics_fn=process_metrics_fn)
      write_metrics(metrics, global_step, summary_dir)
      if (max_evaluation_global_step and
          global_step >= max_evaluation_global_step):
        tf.logging.info('Finished evaluation!')
        break
    number_of_evaluations += 1

    if (max_number_of_evaluations and
        number_of_evaluations >= max_number_of_evaluations):
      tf.logging.info('Finished evaluation!')
      break
    time_to_next_eval = start + eval_interval_secs - time.time()
    if time_to_next_eval > 0:
      time.sleep(time_to_next_eval)

  return metrics


def _scale_box_to_absolute(args):
  boxes, image_shape = args
  return box_list_ops.to_absolute_coordinates(
      box_list.BoxList(boxes), image_shape[0], image_shape[1]).get()


def _resize_detection_masks(args):
  detection_boxes, detection_masks, image_shape = args
  detection_masks_reframed = ops.reframe_box_masks_to_image_masks(
      detection_masks, detection_boxes, image_shape[0], image_shape[1])
  return tf.cast(tf.greater(detection_masks_reframed, 0.5), tf.uint8)


def _resize_groundtruth_masks(args):
  """Resizes groundgtruth masks to the original image size."""
  mask, true_image_shape, original_image_shape = args
  true_height = true_image_shape[0]
  true_width = true_image_shape[1]
  mask = mask[:, :true_height, :true_width]
  mask = tf.expand_dims(mask, 3)
  mask = tf.image.resize_images(
      mask,
      original_image_shape,
      method=tf.image.ResizeMethod.NEAREST_NEIGHBOR,
      align_corners=True)
  return tf.cast(tf.squeeze(mask, 3), tf.uint8)


def _scale_keypoint_to_absolute(args):
  keypoints, image_shape = args
  return keypoint_ops.scale(keypoints, image_shape[0], image_shape[1])


def result_dict_for_single_example(image,
                                   key,
                                   detections,
                                   groundtruth=None,
                                   class_agnostic=False,
                                   scale_to_absolute=False):
  """Merges all detection and groundtruth information for a single example.

  Note that evaluation tools require classes that are 1-indexed, and so this
  function performs the offset. If `class_agnostic` is True, all output classes
  have label 1.

  Args:
    image: A single 4D uint8 image tensor of shape [1, H, W, C].
    key: A single string tensor identifying the image.
    detections: A dictionary of detections, returned from
      DetectionModel.postprocess().
    groundtruth: (Optional) Dictionary of groundtruth items, with fields:
      'groundtruth_boxes': [num_boxes, 4] float32 tensor of boxes, in
        normalized coordinates.
      'groundtruth_classes': [num_boxes] int64 tensor of 1-indexed classes.
      'groundtruth_area': [num_boxes] float32 tensor of bbox area. (Optional)
      'groundtruth_is_crowd': [num_boxes] int64 tensor. (Optional)
      'groundtruth_difficult': [num_boxes] int64 tensor. (Optional)
      'groundtruth_group_of': [num_boxes] int64 tensor. (Optional)
      'groundtruth_instance_masks': 3D int64 tensor of instance masks
        (Optional).
      'groundtruth_keypoints': [num_boxes, num_keypoints, 2] float32 tensor with
        keypoints (Optional).
    class_agnostic: Boolean indicating whether the detections are class-agnostic
      (i.e. binary). Default False.
    scale_to_absolute: Boolean indicating whether boxes and keypoints should be
      scaled to absolute coordinates. Note that for IoU based evaluations, it
      does not matter whether boxes are expressed in absolute or relative
      coordinates. Default False.

  Returns:
    A dictionary with:
    'original_image': A [1, H, W, C] uint8 image tensor.
    'key': A string tensor with image identifier.
    'detection_boxes': [max_detections, 4] float32 tensor of boxes, in
      normalized or absolute coordinates, depending on the value of
      `scale_to_absolute`.
    'detection_scores': [max_detections] float32 tensor of scores.
    'detection_classes': [max_detections] int64 tensor of 1-indexed classes.
    'detection_masks': [max_detections, H, W] float32 tensor of binarized
      masks, reframed to full image masks.
    'groundtruth_boxes': [num_boxes, 4] float32 tensor of boxes, in
      normalized or absolute coordinates, depending on the value of
      `scale_to_absolute`. (Optional)
    'groundtruth_classes': [num_boxes] int64 tensor of 1-indexed classes.
      (Optional)
    'groundtruth_area': [num_boxes] float32 tensor of bbox area. (Optional)
    'groundtruth_is_crowd': [num_boxes] int64 tensor. (Optional)
    'groundtruth_difficult': [num_boxes] int64 tensor. (Optional)
    'groundtruth_group_of': [num_boxes] int64 tensor. (Optional)
    'groundtruth_instance_masks': 3D int64 tensor of instance masks
      (Optional).
    'groundtruth_keypoints': [num_boxes, num_keypoints, 2] float32 tensor with
      keypoints (Optional).
  """

  if groundtruth:
    max_gt_boxes = tf.shape(
        groundtruth[fields.InputDataFields.groundtruth_boxes])[0]
    for gt_key in groundtruth:
      # expand groundtruth dict along the batch dimension.
      groundtruth[gt_key] = tf.expand_dims(groundtruth[gt_key], 0)

  for detection_key in detections:
    detections[detection_key] = tf.expand_dims(
        detections[detection_key][0], axis=0)

  batched_output_dict = result_dict_for_batched_example(
      image,
      tf.expand_dims(key, 0),
      detections,
      groundtruth,
      class_agnostic,
      scale_to_absolute,
      max_gt_boxes=max_gt_boxes)

  exclude_keys = [
      fields.InputDataFields.original_image,
      fields.DetectionResultFields.num_detections,
      fields.InputDataFields.num_groundtruth_boxes
  ]

  output_dict = {
      fields.InputDataFields.original_image:
          batched_output_dict[fields.InputDataFields.original_image]
  }

  for key in batched_output_dict:
    # remove the batch dimension.
    if key not in exclude_keys:
      output_dict[key] = tf.squeeze(batched_output_dict[key], 0)
  return output_dict


def result_dict_for_batched_example(images,
                                    keys,
                                    detections,
                                    groundtruth=None,
                                    class_agnostic=False,
                                    scale_to_absolute=False,
                                    original_image_spatial_shapes=None,
                                    true_image_shapes=None,
                                    max_gt_boxes=None):
  """Merges all detection and groundtruth information for a single example.

  Note that evaluation tools require classes that are 1-indexed, and so this
  function performs the offset. If `class_agnostic` is True, all output classes
  have label 1.
  The groundtruth coordinates of boxes/keypoints in 'groundtruth' dictionary are
  normalized relative to the (potentially padded) input image, while the
  coordinates in 'detection' dictionary are normalized relative to the true
  image shape.

  Args:
    images: A single 4D uint8 image tensor of shape [batch_size, H, W, C].
    keys: A [batch_size] string/int tensor with image identifier.
    detections: A dictionary of detections, returned from
      DetectionModel.postprocess().
    groundtruth: (Optional) Dictionary of groundtruth items, with fields:
      'groundtruth_boxes': [batch_size, max_number_of_boxes, 4] float32 tensor
        of boxes, in normalized coordinates.
      'groundtruth_classes':  [batch_size, max_number_of_boxes] int64 tensor of
        1-indexed classes.
      'groundtruth_area': [batch_size, max_number_of_boxes] float32 tensor of
        bbox area. (Optional)
      'groundtruth_is_crowd':[batch_size, max_number_of_boxes] int64
        tensor. (Optional)
      'groundtruth_difficult': [batch_size, max_number_of_boxes] int64
        tensor. (Optional)
      'groundtruth_group_of': [batch_size, max_number_of_boxes] int64
        tensor. (Optional)
      'groundtruth_instance_masks': 4D int64 tensor of instance
        masks (Optional).
      'groundtruth_keypoints': [batch_size, max_number_of_boxes, num_keypoints,
        2] float32 tensor with keypoints (Optional).
      'groundtruth_keypoint_visibilities': [batch_size, max_number_of_boxes,
        num_keypoints] bool tensor with keypoint visibilities (Optional).
      'groundtruth_labeled_classes': [batch_size, num_classes] int64
        tensor of 1-indexed classes. (Optional)
    class_agnostic: Boolean indicating whether the detections are class-agnostic
      (i.e. binary). Default False.
    scale_to_absolute: Boolean indicating whether boxes and keypoints should be
      scaled to absolute coordinates. Note that for IoU based evaluations, it
      does not matter whether boxes are expressed in absolute or relative
      coordinates. Default False.
    original_image_spatial_shapes: A 2D int32 tensor of shape [batch_size, 2]
      used to resize the image. When set to None, the image size is retained.
    true_image_shapes: A 2D int32 tensor of shape [batch_size, 3]
      containing the size of the unpadded original_image.
    max_gt_boxes: [batch_size] tensor representing the maximum number of
      groundtruth boxes to pad.

  Returns:
    A dictionary with:
    'original_image': A [batch_size, H, W, C] uint8 image tensor.
    'original_image_spatial_shape': A [batch_size, 2] tensor containing the
      original image sizes.
    'true_image_shape': A [batch_size, 3] tensor containing the size of
      the unpadded original_image.
    'key': A [batch_size] string tensor with image identifier.
    'detection_boxes': [batch_size, max_detections, 4] float32 tensor of boxes,
      in normalized or absolute coordinates, depending on the value of
      `scale_to_absolute`.
    'detection_scores': [batch_size, max_detections] float32 tensor of scores.
    'detection_classes': [batch_size, max_detections] int64 tensor of 1-indexed
      classes.
    'detection_masks': [batch_size, max_detections, H, W] float32 tensor of
      binarized masks, reframed to full image masks. (Optional)
    'detection_keypoints': [batch_size, max_detections, num_keypoints, 2]
      float32 tensor containing keypoint coordinates. (Optional)
    'detection_keypoint_scores': [batch_size, max_detections, num_keypoints]
      float32 tensor containing keypoint scores. (Optional)
    'num_detections': [batch_size] int64 tensor containing number of valid
      detections.
    'groundtruth_boxes': [batch_size, num_boxes, 4] float32 tensor of boxes, in
      normalized or absolute coordinates, depending on the value of
      `scale_to_absolute`. (Optional)
    'groundtruth_classes': [batch_size, num_boxes] int64 tensor of 1-indexed
      classes. (Optional)
    'groundtruth_area': [batch_size, num_boxes] float32 tensor of bbox
      area. (Optional)
    'groundtruth_is_crowd': [batch_size, num_boxes] int64 tensor. (Optional)
    'groundtruth_difficult': [batch_size, num_boxes] int64 tensor. (Optional)
    'groundtruth_group_of': [batch_size, num_boxes] int64 tensor. (Optional)
    'groundtruth_instance_masks': 4D int64 tensor of instance masks
      (Optional).
    'groundtruth_keypoints': [batch_size, num_boxes, num_keypoints, 2] float32
      tensor with keypoints (Optional).
    'groundtruth_keypoint_visibilities': [batch_size, num_boxes, num_keypoints]
      bool tensor with keypoint visibilities (Optional).
    'groundtruth_labeled_classes': [batch_size, num_classes]  int64 tensor
      of 1-indexed classes. (Optional)
    'num_groundtruth_boxes': [batch_size] tensor containing the maximum number
      of groundtruth boxes per image.

  Raises:
    ValueError: if original_image_spatial_shape is not 2D int32 tensor of shape
      [2].
    ValueError: if true_image_shapes is not 2D int32 tensor of shape
      [3].
  """
  label_id_offset = 1  # Applying label id offset (b/63711816)

  input_data_fields = fields.InputDataFields
  if original_image_spatial_shapes is None:
    original_image_spatial_shapes = tf.tile(
        tf.expand_dims(tf.shape(images)[1:3], axis=0),
        multiples=[tf.shape(images)[0], 1])
  else:
    if (len(original_image_spatial_shapes.shape) != 2 and
        original_image_spatial_shapes.shape[1] != 2):
      raise ValueError(
          '`original_image_spatial_shape` should be a 2D tensor of shape '
          '[batch_size, 2].')

  if true_image_shapes is None:
    true_image_shapes = tf.tile(
        tf.expand_dims(tf.shape(images)[1:4], axis=0),
        multiples=[tf.shape(images)[0], 1])
  else:
    if (len(true_image_shapes.shape) != 2
        and true_image_shapes.shape[1] != 3):
      raise ValueError('`true_image_shapes` should be a 2D tensor of '
                       'shape [batch_size, 3].')

  output_dict = {
      input_data_fields.original_image:
          images,
      input_data_fields.key:
          keys,
      input_data_fields.original_image_spatial_shape: (
          original_image_spatial_shapes),
      input_data_fields.true_image_shape:
          true_image_shapes
  }

  detection_fields = fields.DetectionResultFields
  detection_boxes = detections[detection_fields.detection_boxes]
  detection_scores = detections[detection_fields.detection_scores]
  num_detections = tf.cast(detections[detection_fields.num_detections],
                           dtype=tf.int32)

  if class_agnostic:
    detection_classes = tf.ones_like(detection_scores, dtype=tf.int64)
  else:
    detection_classes = (
        tf.to_int64(detections[detection_fields.detection_classes]) +
        label_id_offset)

  if scale_to_absolute:
    output_dict[detection_fields.detection_boxes] = (
        shape_utils.static_or_dynamic_map_fn(
            _scale_box_to_absolute,
            elems=[detection_boxes, original_image_spatial_shapes],
            dtype=tf.float32))
  else:
    output_dict[detection_fields.detection_boxes] = detection_boxes
  output_dict[detection_fields.detection_classes] = detection_classes
  output_dict[detection_fields.detection_scores] = detection_scores
  output_dict[detection_fields.num_detections] = num_detections

  if detection_fields.detection_masks in detections:
    detection_masks = detections[detection_fields.detection_masks]
    # TODO(rathodv): This should be done in model's postprocess
    # function ideally.
    output_dict[detection_fields.detection_masks] = (
        shape_utils.static_or_dynamic_map_fn(
            _resize_detection_masks,
            elems=[detection_boxes, detection_masks,
                   original_image_spatial_shapes],
            dtype=tf.uint8))

  if detection_fields.detection_keypoints in detections:
    detection_keypoints = detections[detection_fields.detection_keypoints]
    output_dict[detection_fields.detection_keypoints] = detection_keypoints
    if scale_to_absolute:
      output_dict[detection_fields.detection_keypoints] = (
          shape_utils.static_or_dynamic_map_fn(
              _scale_keypoint_to_absolute,
              elems=[detection_keypoints, original_image_spatial_shapes],
              dtype=tf.float32))
    if detection_fields.detection_keypoint_scores in detections:
      output_dict[detection_fields.detection_keypoint_scores] = detections[
          detection_fields.detection_keypoint_scores]
    else:
      output_dict[detection_fields.detection_keypoint_scores] = tf.ones_like(
          detections[detection_fields.detection_keypoints][:, :, :, 0])

  if groundtruth:
    if max_gt_boxes is None:
      if input_data_fields.num_groundtruth_boxes in groundtruth:
        max_gt_boxes = groundtruth[input_data_fields.num_groundtruth_boxes]
      else:
        raise ValueError(
            'max_gt_boxes must be provided when processing batched examples.')

    if input_data_fields.groundtruth_instance_masks in groundtruth:
      masks = groundtruth[input_data_fields.groundtruth_instance_masks]
      groundtruth[input_data_fields.groundtruth_instance_masks] = (
          shape_utils.static_or_dynamic_map_fn(
              _resize_groundtruth_masks,
              elems=[masks, true_image_shapes, original_image_spatial_shapes],
              dtype=tf.uint8))

    output_dict.update(groundtruth)

    image_shape = tf.cast(tf.shape(images), tf.float32)
    image_height, image_width = image_shape[1], image_shape[2]

    def _scale_box_to_normalized_true_image(args):
      """Scale the box coordinates to be relative to the true image shape."""
      boxes, true_image_shape = args
      true_image_shape = tf.cast(true_image_shape, tf.float32)
      true_height, true_width = true_image_shape[0], true_image_shape[1]
      normalized_window = tf.stack([0.0, 0.0, true_height / image_height,
                                    true_width / image_width])
      return box_list_ops.change_coordinate_frame(
          box_list.BoxList(boxes), normalized_window).get()

    groundtruth_boxes = groundtruth[input_data_fields.groundtruth_boxes]
    groundtruth_boxes = shape_utils.static_or_dynamic_map_fn(
        _scale_box_to_normalized_true_image,
        elems=[groundtruth_boxes, true_image_shapes], dtype=tf.float32)
    output_dict[input_data_fields.groundtruth_boxes] = groundtruth_boxes

    if input_data_fields.groundtruth_keypoints in groundtruth:
      # If groundtruth_keypoints is in the groundtruth dictionary. Update the
      # coordinates to conform with the true image shape.
      def _scale_keypoints_to_normalized_true_image(args):
        """Scale the box coordinates to be relative to the true image shape."""
        keypoints, true_image_shape = args
        true_image_shape = tf.cast(true_image_shape, tf.float32)
        true_height, true_width = true_image_shape[0], true_image_shape[1]
        normalized_window = tf.stack(
            [0.0, 0.0, true_height / image_height, true_width / image_width])
        return keypoint_ops.change_coordinate_frame(keypoints,
                                                    normalized_window)

      groundtruth_keypoints = groundtruth[
          input_data_fields.groundtruth_keypoints]
      groundtruth_keypoints = shape_utils.static_or_dynamic_map_fn(
          _scale_keypoints_to_normalized_true_image,
          elems=[groundtruth_keypoints, true_image_shapes],
          dtype=tf.float32)
      output_dict[
          input_data_fields.groundtruth_keypoints] = groundtruth_keypoints

    if scale_to_absolute:
      groundtruth_boxes = output_dict[input_data_fields.groundtruth_boxes]
      output_dict[input_data_fields.groundtruth_boxes] = (
          shape_utils.static_or_dynamic_map_fn(
              _scale_box_to_absolute,
              elems=[groundtruth_boxes, original_image_spatial_shapes],
              dtype=tf.float32))
      if input_data_fields.groundtruth_keypoints in groundtruth:
        groundtruth_keypoints = output_dict[
            input_data_fields.groundtruth_keypoints]
        output_dict[input_data_fields.groundtruth_keypoints] = (
            shape_utils.static_or_dynamic_map_fn(
                _scale_keypoint_to_absolute,
                elems=[groundtruth_keypoints, original_image_spatial_shapes],
                dtype=tf.float32))

    # For class-agnostic models, groundtruth classes all become 1.
    if class_agnostic:
      groundtruth_classes = groundtruth[input_data_fields.groundtruth_classes]
      groundtruth_classes = tf.ones_like(groundtruth_classes, dtype=tf.int64)
      output_dict[input_data_fields.groundtruth_classes] = groundtruth_classes

    output_dict[input_data_fields.num_groundtruth_boxes] = max_gt_boxes

  return output_dict


def get_evaluators(eval_config, categories, evaluator_options=None):
  """Returns the evaluator class according to eval_config, valid for categories.

  Args:
    eval_config: An `eval_pb2.EvalConfig`.
    categories: A list of dicts, each of which has the following keys -
        'id': (required) an integer id uniquely identifying this category.
        'name': (required) string representing category name e.g., 'cat', 'dog'.
        'keypoints': (optional) dict mapping this category's keypoints to unique
          ids.
    evaluator_options: A dictionary of metric names (see
      EVAL_METRICS_CLASS_DICT) to `DetectionEvaluator` initialization
      keyword arguments. For example:
      evalator_options = {
        'coco_detection_metrics': {'include_metrics_per_category': True}
      }

  Returns:
    An list of instances of DetectionEvaluator.

  Raises:
    ValueError: if metric is not in the metric class dictionary.
  """
  evaluator_options = evaluator_options or {}
  eval_metric_fn_keys = eval_config.metrics_set
  if not eval_metric_fn_keys:
    eval_metric_fn_keys = [EVAL_DEFAULT_METRIC]
  evaluators_list = []
  for eval_metric_fn_key in eval_metric_fn_keys:
    if eval_metric_fn_key not in EVAL_METRICS_CLASS_DICT:
      raise ValueError('Metric not found: {}'.format(eval_metric_fn_key))
    kwargs_dict = (evaluator_options[eval_metric_fn_key] if eval_metric_fn_key
                   in evaluator_options else {})
    evaluators_list.append(EVAL_METRICS_CLASS_DICT[eval_metric_fn_key](
        categories,
        **kwargs_dict))

  if isinstance(eval_config, eval_pb2.EvalConfig):
    parameterized_metrics = eval_config.parameterized_metric
    for parameterized_metric in parameterized_metrics:
      assert parameterized_metric.HasField('parameterized_metric')
      if parameterized_metric.WhichOneof(
          'parameterized_metric') == EVAL_KEYPOINT_METRIC:
        keypoint_metrics = parameterized_metric.coco_keypoint_metrics
        # Create category to keypoints mapping dict.
        category_keypoints = {}
        class_label = keypoint_metrics.class_label
        category = None
        for cat in categories:
          if cat['name'] == class_label:
            category = cat
            break
        if not category:
          continue
        keypoints_for_this_class = category['keypoints']
        category_keypoints = [{
            'id': keypoints_for_this_class[kp_name], 'name': kp_name
        } for kp_name in keypoints_for_this_class]
        # Create keypoint evaluator for this category.
        evaluators_list.append(EVAL_METRICS_CLASS_DICT[EVAL_KEYPOINT_METRIC](
            category['id'], category_keypoints, class_label,
            keypoint_metrics.keypoint_label_to_sigmas))
  return evaluators_list


def get_eval_metric_ops_for_evaluators(eval_config,
                                       categories,
                                       eval_dict):
  """Returns eval metrics ops to use with `tf.estimator.EstimatorSpec`.

  Args:
    eval_config: An `eval_pb2.EvalConfig`.
    categories: A list of dicts, each of which has the following keys -
        'id': (required) an integer id uniquely identifying this category.
        'name': (required) string representing category name e.g., 'cat', 'dog'.
    eval_dict: An evaluation dictionary, returned from
      result_dict_for_single_example().

  Returns:
    A dictionary of metric names to tuple of value_op and update_op that can be
    used as eval metric ops in tf.EstimatorSpec.
  """
  eval_metric_ops = {}
  evaluator_options = evaluator_options_from_eval_config(eval_config)
  evaluators_list = get_evaluators(eval_config, categories, evaluator_options)
  for evaluator in evaluators_list:
    eval_metric_ops.update(evaluator.get_estimator_eval_metric_ops(
        eval_dict))
  return eval_metric_ops


def evaluator_options_from_eval_config(eval_config):
  """Produces a dictionary of evaluation options for each eval metric.

  Args:
    eval_config: An `eval_pb2.EvalConfig`.

  Returns:
    evaluator_options: A dictionary of metric names (see
      EVAL_METRICS_CLASS_DICT) to `DetectionEvaluator` initialization
      keyword arguments. For example:
      evalator_options = {
        'coco_detection_metrics': {'include_metrics_per_category': True}
      }
  """
  eval_metric_fn_keys = eval_config.metrics_set
  evaluator_options = {}
  for eval_metric_fn_key in eval_metric_fn_keys:
    if eval_metric_fn_key in ('coco_detection_metrics', 'coco_mask_metrics'):
      evaluator_options[eval_metric_fn_key] = {
          'include_metrics_per_category': (
              eval_config.include_metrics_per_category)
      }
    elif eval_metric_fn_key == 'precision_at_recall_detection_metrics':
      evaluator_options[eval_metric_fn_key] = {
          'recall_lower_bound': (eval_config.recall_lower_bound),
          'recall_upper_bound': (eval_config.recall_upper_bound)
      }
  return evaluator_options