# Copyright 2022 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.
"""Tests for image_searcher."""
import enum
from absl.testing import parameterized
import tensorflow as tf
from tensorflow_lite_support.python.task.core import base_options as base_options_module
from tensorflow_lite_support.python.task.processor.proto import bounding_box_pb2
from tensorflow_lite_support.python.task.processor.proto import embedding_options_pb2
from tensorflow_lite_support.python.task.processor.proto import search_options_pb2
from tensorflow_lite_support.python.task.processor.proto import search_result_pb2
from tensorflow_lite_support.python.task.vision import image_searcher
from tensorflow_lite_support.python.task.vision.core import tensor_image
from tensorflow_lite_support.python.test import test_util
_BaseOptions = base_options_module.BaseOptions
_EmbeddingOptions = embedding_options_pb2.EmbeddingOptions
_SearchOptions = search_options_pb2.SearchOptions
_SearchResult = search_result_pb2.SearchResult
_NearestNeighbor = search_result_pb2.NearestNeighbor
_ImageSearcher = image_searcher.ImageSearcher
_ImageSearcherOptions = image_searcher.ImageSearcherOptions
_MOBILENET_EMBEDDER_MODEL = 'mobilenet_v3_small_100_224_embedder.tflite'
_MOBILENET_SEARCHER_MODEL = 'mobilenet_v3_small_100_224_searcher.tflite'
_MOBILENET_INDEX = 'searcher_index.ldb'
_EXPECTED_MOBILENET_DEFAULT_OPTIONS_SEARCH_RESULT = _SearchResult(
nearest_neighbors=[
_NearestNeighbor(metadata=bytearray(b'burger'), distance=200.798508),
_NearestNeighbor(metadata=bytearray(b'car'), distance=228.445480),
_NearestNeighbor(metadata=bytearray(b'bird'), distance=230.091507),
_NearestNeighbor(metadata=bytearray(b'dog'), distance=231.857605),
_NearestNeighbor(metadata=bytearray(b'cat'), distance=232.290115)
])
_IMAGE_FILE = 'burger.jpg'
_MAX_RESULTS = 2
class ModelFileType(enum.Enum):
FILE_CONTENT = 1
FILE_NAME = 2
class IndexFileType(enum.Enum):
NONE = 1
FILE_CONTENT = 2
FILE_NAME = 3
class ImageSearcherTest(parameterized.TestCase, tf.test.TestCase):
def setUp(self):
super().setUp()
self.test_image_path = test_util.get_test_data_path(_IMAGE_FILE)
self.embedder_model_path = test_util.get_test_data_path(
_MOBILENET_EMBEDDER_MODEL)
self.searcher_model_path = test_util.get_test_data_path(
_MOBILENET_SEARCHER_MODEL)
self.index_path = test_util.get_test_data_path(_MOBILENET_INDEX)
def test_create_from_file_succeeds_with_valid_embedder_and_index_paths(self):
# Creates with default option and valid model and index files successfully.
searcher = _ImageSearcher.create_from_file(self.embedder_model_path,
self.index_path)
self.assertIsInstance(searcher, _ImageSearcher)
def test_create_from_file_succeeds_with_valid_searcher_path(self):
# Creates with default option and valid searcher model.
searcher = _ImageSearcher.create_from_file(self.searcher_model_path)
self.assertIsInstance(searcher, _ImageSearcher)
def test_create_from_options_succeeds_with_valid_embedder_and_index_paths(
self):
options = _ImageSearcherOptions(
base_options=_BaseOptions(file_name=self.embedder_model_path),
search_options=_SearchOptions(index_file_name=self.index_path))
searcher = _ImageSearcher.create_from_options(options)
self.assertIsInstance(searcher, _ImageSearcher)
def test_create_from_options_succeeds_with_valid_searcher_path(self):
options = _ImageSearcherOptions(
base_options=_BaseOptions(file_name=self.searcher_model_path),
search_options=_SearchOptions())
searcher = _ImageSearcher.create_from_options(options)
self.assertIsInstance(searcher, _ImageSearcher)
def test_create_from_options_succeeds_with_valid_embedder_content(self):
# Creates with options containing model content successfully.
with open(self.embedder_model_path, 'rb') as f:
options = _ImageSearcherOptions(
base_options=_BaseOptions(file_content=f.read()),
search_options=_SearchOptions(index_file_name=self.index_path))
searcher = _ImageSearcher.create_from_options(options)
self.assertIsInstance(searcher, _ImageSearcher)
def test_create_from_options_succeeds_with_valid_searcher_content(self):
# Creates with options containing model content successfully.
with open(self.searcher_model_path, 'rb') as f:
options = _ImageSearcherOptions(
base_options=_BaseOptions(file_content=f.read()),
search_options=_SearchOptions())
searcher = _ImageSearcher.create_from_options(options)
self.assertIsInstance(searcher, _ImageSearcher)
def test_create_from_options_succeeds_with_valid_index_content(self):
# Creates with options containing index content successfully.
with open(self.index_path, 'rb') as f:
options = _ImageSearcherOptions(
base_options=_BaseOptions(file_name=self.embedder_model_path),
search_options=_SearchOptions(index_file_content=f.read()))
searcher = _ImageSearcher.create_from_options(options)
self.assertIsInstance(searcher, _ImageSearcher)
def test_create_from_options_fails_with_invalid_index_path(self):
# Invalid index path.
with self.assertRaisesRegex(
ValueError,
r'Unable to find index file: SearchOptions.index_file is not set and '
r'no AssociatedFile with type SCANN_INDEX_FILE could be found in the '
r'output tensor metadata.'):
options = _ImageSearcherOptions(
base_options=_BaseOptions(file_name=self.embedder_model_path))
_ImageSearcher.create_from_options(options)
def test_create_from_options_fails_with_invalid_model_path(self):
# Invalid empty model path.
with self.assertRaisesRegex(
ValueError,
r"ExternalFile must specify at least one of 'file_content', "
r"'file_name' or 'file_descriptor_meta'."):
options = _ImageSearcherOptions(
base_options=_BaseOptions(file_name=''),
search_options=_SearchOptions(index_file_name=self.index_path))
_ImageSearcher.create_from_options(options)
def test_create_from_options_fails_with_invalid_quantization(self):
# Invalid quantization option.
with self.assertRaisesRegex(
ValueError,
r'Setting EmbeddingOptions.quantize = true is not allowed in '
r'searchers.'):
options = _ImageSearcherOptions(
base_options=_BaseOptions(file_name=self.embedder_model_path),
embedding_options=_EmbeddingOptions(quantize=True),
search_options=_SearchOptions(index_file_name=self.index_path))
_ImageSearcher.create_from_options(options)
def test_create_from_options_fails_with_invalid_max_results(self):
# Invalid max results option.
with self.assertRaisesRegex(
ValueError, r'SearchOptions.max_results must be > 0, found -1.'):
options = _ImageSearcherOptions(
base_options=_BaseOptions(file_name=self.embedder_model_path),
search_options=_SearchOptions(
index_file_name=self.index_path, max_results=-1))
_ImageSearcher.create_from_options(options)
def test_search_with_default_options(self):
# Create searcher.
searcher = _ImageSearcher.create_from_file(self.embedder_model_path,
self.index_path)
# Loads image.
image = tensor_image.TensorImage.create_from_file(self.test_image_path)
# Perform image search.
image_search_result = searcher.search(image)
self.assertProtoEquals(
image_search_result.to_pb2(),
_EXPECTED_MOBILENET_DEFAULT_OPTIONS_SEARCH_RESULT.to_pb2())
@parameterized.parameters(
(_MOBILENET_EMBEDDER_MODEL, ModelFileType.FILE_NAME,
IndexFileType.FILE_NAME),
(_MOBILENET_EMBEDDER_MODEL, ModelFileType.FILE_CONTENT,
IndexFileType.FILE_NAME),
(_MOBILENET_EMBEDDER_MODEL, ModelFileType.FILE_NAME,
IndexFileType.FILE_CONTENT),
(_MOBILENET_EMBEDDER_MODEL, ModelFileType.FILE_CONTENT,
IndexFileType.FILE_CONTENT),
(_MOBILENET_SEARCHER_MODEL, ModelFileType.FILE_NAME, IndexFileType.NONE),
(_MOBILENET_SEARCHER_MODEL, ModelFileType.FILE_CONTENT,
IndexFileType.NONE),
)
def test_search(self, model_name, model_file_type, index_file_type):
# Create BaseOptions.
model_path = test_util.get_test_data_path(model_name)
if model_file_type is ModelFileType.FILE_NAME:
base_options = _BaseOptions(file_name=model_path)
elif model_file_type is ModelFileType.FILE_CONTENT:
with open(model_path, 'rb') as f:
model_content = f.read()
base_options = _BaseOptions(file_content=model_content)
else:
# Should never happen
raise ValueError('model_file_type is invalid.')
# Create SearchOptions.
if index_file_type is IndexFileType.NONE:
search_options = _SearchOptions()
else:
index_path = test_util.get_test_data_path(_MOBILENET_INDEX)
if index_file_type is IndexFileType.FILE_NAME:
search_options = _SearchOptions(index_file_name=index_path)
elif index_file_type is IndexFileType.FILE_CONTENT:
with open(index_path, 'rb') as f:
index_content = f.read()
search_options = _SearchOptions(index_file_content=index_content)
else:
# Should never happen
raise ValueError('index_file_type is invalid.')
# Create searcher.
options = _ImageSearcherOptions(
base_options, _EmbeddingOptions(l2_normalize=True, quantize=False),
search_options)
searcher = _ImageSearcher.create_from_options(options)
# Loads image.
image = tensor_image.TensorImage.create_from_file(self.test_image_path)
# Perform image search.
image_search_result = searcher.search(image)
# Expected results.
expected_search_result = _SearchResult(nearest_neighbors=[
_NearestNeighbor(metadata=bytearray(b'burger'), distance=-0.0),
_NearestNeighbor(metadata=bytearray(b'car'), distance=1.822435),
_NearestNeighbor(metadata=bytearray(b'bird'), distance=1.930939),
_NearestNeighbor(metadata=bytearray(b'dog'), distance=2.047355),
_NearestNeighbor(metadata=bytearray(b'cat'), distance=2.075868)
])
# Comparing results.
self.assertProtoEquals(image_search_result.to_pb2(),
expected_search_result.to_pb2())
# Get user info and compare values.
self.assertEqual(searcher.get_user_info(), 'userinfo')
def test_search_with_bounding_box(self):
# Create searcher.
searcher = _ImageSearcher.create_from_file(self.embedder_model_path,
self.index_path)
# Loads image.
image = tensor_image.TensorImage.create_from_file(self.test_image_path)
# Bounding box in "burger.jpg" corresponding to "burger_crop.jpg".
bounding_box = bounding_box_pb2.BoundingBox(
origin_x=0, origin_y=0, width=400, height=325)
# Perform image search.
image_search_result = searcher.search(image, bounding_box)
# Expected results.
expected_search_result = _SearchResult(nearest_neighbors=[
_NearestNeighbor(metadata=bytearray(b'burger'), distance=184.85214),
_NearestNeighbor(metadata=bytearray(b'car'), distance=209.32019),
_NearestNeighbor(metadata=bytearray(b'bird'), distance=211.43195),
_NearestNeighbor(metadata=bytearray(b'dog'), distance=212.77237),
_NearestNeighbor(metadata=bytearray(b'cat'), distance=212.8553)
])
# Comparing results.
self.assertProtoEquals(image_search_result.to_pb2(),
expected_search_result.to_pb2())
# Get user info and compare values.
self.assertEqual(searcher.get_user_info(), 'userinfo')
def test_max_results_option(self):
# Create searcher.
base_options = _BaseOptions(file_name=self.embedder_model_path)
search_options = _SearchOptions(
index_file_name=self.index_path, max_results=_MAX_RESULTS)
options = _ImageSearcherOptions(base_options,
_EmbeddingOptions(l2_normalize=True),
search_options)
searcher = _ImageSearcher.create_from_options(options)
# Loads image.
image = tensor_image.TensorImage.create_from_file(self.test_image_path)
# Perform image search.
image_search_result = searcher.search(image)
nearest_neighbors = image_search_result.nearest_neighbors
self.assertLessEqual(
len(nearest_neighbors), _MAX_RESULTS, 'Too many results returned.')
if __name__ == '__main__':
tf.test.main()
Codebase Browser
chromium