Test your AI Agents with the all-new Agent to Agent Testing Platform.Learn More

How to use classification method in assertpy

Best Python code snippet using assertpy_python

losses_builder_test.py

Source:losses_builder_test.py

1# Copyright 2017 The TensorFlow Authors. All Rights Reserved.2#3# Licensed under the Apache License, Version 2.0 (the "License");4# you may not use this file except in compliance with the License.5# You may obtain a copy of the License at6#7#     http://www.apache.org/licenses/LICENSE-2.08#9# Unless required by applicable law or agreed to in writing, software10# distributed under the License is distributed on an "AS IS" BASIS,11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.12# See the License for the specific language governing permissions and13# limitations under the License.14# ==============================================================================15"""Tests for losses_builder."""16import tensorflow as tf17from google.protobuf import text_format18from object_detection.builders import losses_builder19from object_detection.core import losses20from object_detection.protos import losses_pb221from object_detection.utils import ops22class LocalizationLossBuilderTest(tf.test.TestCase):23  def test_build_weighted_l2_localization_loss(self):24    losses_text_proto = """25      localization_loss {26        weighted_l2 {27        }28      }29      classification_loss {30        weighted_softmax {31        }32      }33    """34    losses_proto = losses_pb2.Loss()35    text_format.Merge(losses_text_proto, losses_proto)36    _, localization_loss, _, _, _, _, _ = losses_builder.build(losses_proto)37    self.assertTrue(isinstance(localization_loss,38                               losses.WeightedL2LocalizationLoss))39  def test_build_weighted_smooth_l1_localization_loss_default_delta(self):40    losses_text_proto = """41      localization_loss {42        weighted_smooth_l1 {43        }44      }45      classification_loss {46        weighted_softmax {47        }48      }49    """50    losses_proto = losses_pb2.Loss()51    text_format.Merge(losses_text_proto, losses_proto)52    _, localization_loss, _, _, _, _, _ = losses_builder.build(losses_proto)53    self.assertTrue(isinstance(localization_loss,54                               losses.WeightedSmoothL1LocalizationLoss))55    self.assertAlmostEqual(localization_loss._delta, 1.0)56  def test_build_weighted_smooth_l1_localization_loss_non_default_delta(self):57    losses_text_proto = """58      localization_loss {59        weighted_smooth_l1 {60          delta: 0.161        }62      }63      classification_loss {64        weighted_softmax {65        }66      }67    """68    losses_proto = losses_pb2.Loss()69    text_format.Merge(losses_text_proto, losses_proto)70    _, localization_loss, _, _, _, _, _ = losses_builder.build(losses_proto)71    self.assertTrue(isinstance(localization_loss,72                               losses.WeightedSmoothL1LocalizationLoss))73    self.assertAlmostEqual(localization_loss._delta, 0.1)74  def test_build_weighted_iou_localization_loss(self):75    losses_text_proto = """76      localization_loss {77        weighted_iou {78        }79      }80      classification_loss {81        weighted_softmax {82        }83      }84    """85    losses_proto = losses_pb2.Loss()86    text_format.Merge(losses_text_proto, losses_proto)87    _, localization_loss, _, _, _, _, _ = losses_builder.build(losses_proto)88    self.assertTrue(isinstance(localization_loss,89                               losses.WeightedIOULocalizationLoss))90  def test_anchorwise_output(self):91    losses_text_proto = """92      localization_loss {93        weighted_smooth_l1 {94        }95      }96      classification_loss {97        weighted_softmax {98        }99      }100    """101    losses_proto = losses_pb2.Loss()102    text_format.Merge(losses_text_proto, losses_proto)103    _, localization_loss, _, _, _, _, _ = losses_builder.build(losses_proto)104    self.assertTrue(isinstance(localization_loss,105                               losses.WeightedSmoothL1LocalizationLoss))106    predictions = tf.constant([[[0.0, 0.0, 1.0, 1.0], [0.0, 0.0, 1.0, 1.0]]])107    targets = tf.constant([[[0.0, 0.0, 1.0, 1.0], [0.0, 0.0, 1.0, 1.0]]])108    weights = tf.constant([[1.0, 1.0]])109    loss = localization_loss(predictions, targets, weights=weights)110    self.assertEqual(loss.shape, [1, 2])111  def test_raise_error_on_empty_localization_config(self):112    losses_text_proto = """113      classification_loss {114        weighted_softmax {115        }116      }117    """118    losses_proto = losses_pb2.Loss()119    text_format.Merge(losses_text_proto, losses_proto)120    with self.assertRaises(ValueError):121      losses_builder._build_localization_loss(losses_proto)122class ClassificationLossBuilderTest(tf.test.TestCase):123  def test_build_weighted_sigmoid_classification_loss(self):124    losses_text_proto = """125      classification_loss {126        weighted_sigmoid {127        }128      }129      localization_loss {130        weighted_l2 {131        }132      }133    """134    losses_proto = losses_pb2.Loss()135    text_format.Merge(losses_text_proto, losses_proto)136    classification_loss, _, _, _, _, _, _ = losses_builder.build(losses_proto)137    self.assertTrue(isinstance(classification_loss,138                               losses.WeightedSigmoidClassificationLoss))139  def test_build_weighted_sigmoid_focal_classification_loss(self):140    losses_text_proto = """141      classification_loss {142        weighted_sigmoid_focal {143        }144      }145      localization_loss {146        weighted_l2 {147        }148      }149    """150    losses_proto = losses_pb2.Loss()151    text_format.Merge(losses_text_proto, losses_proto)152    classification_loss, _, _, _, _, _, _ = losses_builder.build(losses_proto)153    self.assertTrue(isinstance(classification_loss,154                               losses.SigmoidFocalClassificationLoss))155    self.assertAlmostEqual(classification_loss._alpha, None)156    self.assertAlmostEqual(classification_loss._gamma, 2.0)157  def test_build_weighted_sigmoid_focal_loss_non_default(self):158    losses_text_proto = """159      classification_loss {160        weighted_sigmoid_focal {161          alpha: 0.25162          gamma: 3.0163        }164      }165      localization_loss {166        weighted_l2 {167        }168      }169    """170    losses_proto = losses_pb2.Loss()171    text_format.Merge(losses_text_proto, losses_proto)172    classification_loss, _, _, _, _, _, _ = losses_builder.build(losses_proto)173    self.assertTrue(isinstance(classification_loss,174                               losses.SigmoidFocalClassificationLoss))175    self.assertAlmostEqual(classification_loss._alpha, 0.25)176    self.assertAlmostEqual(classification_loss._gamma, 3.0)177  def test_build_weighted_softmax_classification_loss(self):178    losses_text_proto = """179      classification_loss {180        weighted_softmax {181        }182      }183      localization_loss {184        weighted_l2 {185        }186      }187    """188    losses_proto = losses_pb2.Loss()189    text_format.Merge(losses_text_proto, losses_proto)190    classification_loss, _, _, _, _, _, _ = losses_builder.build(losses_proto)191    self.assertTrue(isinstance(classification_loss,192                               losses.WeightedSoftmaxClassificationLoss))193  def test_build_weighted_logits_softmax_classification_loss(self):194    losses_text_proto = """195      classification_loss {196        weighted_logits_softmax {197        }198      }199      localization_loss {200        weighted_l2 {201        }202      }203    """204    losses_proto = losses_pb2.Loss()205    text_format.Merge(losses_text_proto, losses_proto)206    classification_loss, _, _, _, _, _, _ = losses_builder.build(losses_proto)207    self.assertTrue(208        isinstance(classification_loss,209                   losses.WeightedSoftmaxClassificationAgainstLogitsLoss))210  def test_build_weighted_softmax_classification_loss_with_logit_scale(self):211    losses_text_proto = """212      classification_loss {213        weighted_softmax {214          logit_scale: 2.0215        }216      }217      localization_loss {218        weighted_l2 {219        }220      }221    """222    losses_proto = losses_pb2.Loss()223    text_format.Merge(losses_text_proto, losses_proto)224    classification_loss, _, _, _, _, _, _ = losses_builder.build(losses_proto)225    self.assertTrue(isinstance(classification_loss,226                               losses.WeightedSoftmaxClassificationLoss))227  def test_build_bootstrapped_sigmoid_classification_loss(self):228    losses_text_proto = """229      classification_loss {230        bootstrapped_sigmoid {231          alpha: 0.5232        }233      }234      localization_loss {235        weighted_l2 {236        }237      }238    """239    losses_proto = losses_pb2.Loss()240    text_format.Merge(losses_text_proto, losses_proto)241    classification_loss, _, _, _, _, _, _ = losses_builder.build(losses_proto)242    self.assertTrue(isinstance(classification_loss,243                               losses.BootstrappedSigmoidClassificationLoss))244  def test_anchorwise_output(self):245    losses_text_proto = """246      classification_loss {247        weighted_sigmoid {248          anchorwise_output: true249        }250      }251      localization_loss {252        weighted_l2 {253        }254      }255    """256    losses_proto = losses_pb2.Loss()257    text_format.Merge(losses_text_proto, losses_proto)258    classification_loss, _, _, _, _, _, _ = losses_builder.build(losses_proto)259    self.assertTrue(isinstance(classification_loss,260                               losses.WeightedSigmoidClassificationLoss))261    predictions = tf.constant([[[0.0, 1.0, 0.0], [0.0, 0.5, 0.5]]])262    targets = tf.constant([[[0.0, 1.0, 0.0], [0.0, 0.0, 1.0]]])263    weights = tf.constant([[[1.0, 1.0, 1.0], [1.0, 1.0, 1.0]]])264    loss = classification_loss(predictions, targets, weights=weights)265    self.assertEqual(loss.shape, [1, 2, 3])266  def test_raise_error_on_empty_config(self):267    losses_text_proto = """268      localization_loss {269        weighted_l2 {270        }271      }272    """273    losses_proto = losses_pb2.Loss()274    text_format.Merge(losses_text_proto, losses_proto)275    with self.assertRaises(ValueError):276      losses_builder.build(losses_proto)277class HardExampleMinerBuilderTest(tf.test.TestCase):278  def test_do_not_build_hard_example_miner_by_default(self):279    losses_text_proto = """280      localization_loss {281        weighted_l2 {282        }283      }284      classification_loss {285        weighted_softmax {286        }287      }288    """289    losses_proto = losses_pb2.Loss()290    text_format.Merge(losses_text_proto, losses_proto)291    _, _, _, _, hard_example_miner, _, _ = losses_builder.build(losses_proto)292    self.assertEqual(hard_example_miner, None)293  def test_build_hard_example_miner_for_classification_loss(self):294    losses_text_proto = """295      localization_loss {296        weighted_l2 {297        }298      }299      classification_loss {300        weighted_softmax {301        }302      }303      hard_example_miner {304        loss_type: CLASSIFICATION305      }306    """307    losses_proto = losses_pb2.Loss()308    text_format.Merge(losses_text_proto, losses_proto)309    _, _, _, _, hard_example_miner, _, _ = losses_builder.build(losses_proto)310    self.assertTrue(isinstance(hard_example_miner, losses.HardExampleMiner))311    self.assertEqual(hard_example_miner._loss_type, 'cls')312  def test_build_hard_example_miner_for_localization_loss(self):313    losses_text_proto = """314      localization_loss {315        weighted_l2 {316        }317      }318      classification_loss {319        weighted_softmax {320        }321      }322      hard_example_miner {323        loss_type: LOCALIZATION324      }325    """326    losses_proto = losses_pb2.Loss()327    text_format.Merge(losses_text_proto, losses_proto)328    _, _, _, _, hard_example_miner, _, _ = losses_builder.build(losses_proto)329    self.assertTrue(isinstance(hard_example_miner, losses.HardExampleMiner))330    self.assertEqual(hard_example_miner._loss_type, 'loc')331  def test_build_hard_example_miner_with_non_default_values(self):332    losses_text_proto = """333      localization_loss {334        weighted_l2 {335        }336      }337      classification_loss {338        weighted_softmax {339        }340      }341      hard_example_miner {342        num_hard_examples: 32343        iou_threshold: 0.5344        loss_type: LOCALIZATION345        max_negatives_per_positive: 10346        min_negatives_per_image: 3347      }348    """349    losses_proto = losses_pb2.Loss()350    text_format.Merge(losses_text_proto, losses_proto)351    _, _, _, _, hard_example_miner, _, _ = losses_builder.build(losses_proto)352    self.assertTrue(isinstance(hard_example_miner, losses.HardExampleMiner))353    self.assertEqual(hard_example_miner._num_hard_examples, 32)354    self.assertAlmostEqual(hard_example_miner._iou_threshold, 0.5)355    self.assertEqual(hard_example_miner._max_negatives_per_positive, 10)356    self.assertEqual(hard_example_miner._min_negatives_per_image, 3)357class LossBuilderTest(tf.test.TestCase):358  def test_build_all_loss_parameters(self):359    losses_text_proto = """360      localization_loss {361        weighted_l2 {362        }363      }364      classification_loss {365        weighted_softmax {366        }367      }368      hard_example_miner {369      }370      classification_weight: 0.8371      localization_weight: 0.2372    """373    losses_proto = losses_pb2.Loss()374    text_format.Merge(losses_text_proto, losses_proto)375    (classification_loss, localization_loss, classification_weight,376     localization_weight, hard_example_miner, _,377     _) = losses_builder.build(losses_proto)378    self.assertTrue(isinstance(hard_example_miner, losses.HardExampleMiner))379    self.assertTrue(isinstance(classification_loss,380                               losses.WeightedSoftmaxClassificationLoss))381    self.assertTrue(isinstance(localization_loss,382                               losses.WeightedL2LocalizationLoss))383    self.assertAlmostEqual(classification_weight, 0.8)384    self.assertAlmostEqual(localization_weight, 0.2)385  def test_build_expected_sampling(self):386    losses_text_proto = """387      localization_loss {388        weighted_l2 {389        }390      }391      classification_loss {392        weighted_softmax {393        }394      }395      hard_example_miner {396      }397      classification_weight: 0.8398      localization_weight: 0.2399    """400    losses_proto = losses_pb2.Loss()401    text_format.Merge(losses_text_proto, losses_proto)402    (classification_loss, localization_loss, classification_weight,403     localization_weight, hard_example_miner, _,404     _) = losses_builder.build(losses_proto)405    self.assertTrue(isinstance(hard_example_miner, losses.HardExampleMiner))406    self.assertTrue(407        isinstance(classification_loss,408                   losses.WeightedSoftmaxClassificationLoss))409    self.assertTrue(410        isinstance(localization_loss, losses.WeightedL2LocalizationLoss))411    self.assertAlmostEqual(classification_weight, 0.8)412    self.assertAlmostEqual(localization_weight, 0.2)413  def test_build_reweighting_unmatched_anchors(self):414    losses_text_proto = """415      localization_loss {416        weighted_l2 {417        }418      }419      classification_loss {420        weighted_softmax {421        }422      }423      hard_example_miner {424      }425      classification_weight: 0.8426      localization_weight: 0.2427    """428    losses_proto = losses_pb2.Loss()429    text_format.Merge(losses_text_proto, losses_proto)430    (classification_loss, localization_loss, classification_weight,431     localization_weight, hard_example_miner, _,432     _) = losses_builder.build(losses_proto)433    self.assertTrue(isinstance(hard_example_miner, losses.HardExampleMiner))434    self.assertTrue(435        isinstance(classification_loss,436                   losses.WeightedSoftmaxClassificationLoss))437    self.assertTrue(438        isinstance(localization_loss, losses.WeightedL2LocalizationLoss))439    self.assertAlmostEqual(classification_weight, 0.8)440    self.assertAlmostEqual(localization_weight, 0.2)441  def test_raise_error_when_both_focal_loss_and_hard_example_miner(self):442    losses_text_proto = """443      localization_loss {444        weighted_l2 {445        }446      }447      classification_loss {448        weighted_sigmoid_focal {449        }450      }451      hard_example_miner {452      }453      classification_weight: 0.8454      localization_weight: 0.2455    """456    losses_proto = losses_pb2.Loss()457    text_format.Merge(losses_text_proto, losses_proto)458    with self.assertRaises(ValueError):459      losses_builder.build(losses_proto)460class FasterRcnnClassificationLossBuilderTest(tf.test.TestCase):461  def test_build_sigmoid_loss(self):462    losses_text_proto = """463      weighted_sigmoid {464      }465    """466    losses_proto = losses_pb2.ClassificationLoss()467    text_format.Merge(losses_text_proto, losses_proto)468    classification_loss = losses_builder.build_faster_rcnn_classification_loss(469        losses_proto)470    self.assertTrue(isinstance(classification_loss,471                               losses.WeightedSigmoidClassificationLoss))472  def test_build_softmax_loss(self):473    losses_text_proto = """474      weighted_softmax {475      }476    """477    losses_proto = losses_pb2.ClassificationLoss()478    text_format.Merge(losses_text_proto, losses_proto)479    classification_loss = losses_builder.build_faster_rcnn_classification_loss(480        losses_proto)481    self.assertTrue(isinstance(classification_loss,482                               losses.WeightedSoftmaxClassificationLoss))483  def test_build_logits_softmax_loss(self):484    losses_text_proto = """485      weighted_logits_softmax {486      }487    """488    losses_proto = losses_pb2.ClassificationLoss()489    text_format.Merge(losses_text_proto, losses_proto)490    classification_loss = losses_builder.build_faster_rcnn_classification_loss(491        losses_proto)492    self.assertTrue(493        isinstance(classification_loss,494                   losses.WeightedSoftmaxClassificationAgainstLogitsLoss))495  def test_build_sigmoid_focal_loss(self):496    losses_text_proto = """497      weighted_sigmoid_focal {498      }499    """500    losses_proto = losses_pb2.ClassificationLoss()501    text_format.Merge(losses_text_proto, losses_proto)502    classification_loss = losses_builder.build_faster_rcnn_classification_loss(503        losses_proto)504    self.assertTrue(505        isinstance(classification_loss,506                   losses.SigmoidFocalClassificationLoss))507  def test_build_softmax_loss_by_default(self):508    losses_text_proto = """509    """510    losses_proto = losses_pb2.ClassificationLoss()511    text_format.Merge(losses_text_proto, losses_proto)512    classification_loss = losses_builder.build_faster_rcnn_classification_loss(513        losses_proto)514    self.assertTrue(isinstance(classification_loss,515                               losses.WeightedSoftmaxClassificationLoss))516if __name__ == '__main__':...

config.py

Source:config.py

1from constants import *2CLASSIFICATIONS = {3    "catvsDog": {4        "img_size"   : 224,5        "normalize"  : True,6    },7    "faceMaskClassification": {8        "img_size"   : 224,9        "normalize"  : True,10    },11    "genderClassification": {12        "img_size"   : 150,13        "normalize"  : True,14    },15    "emotionClassification": {16        "img_size"   : 48,17        "normalize"  : True,18    },19    "glassesClassification":{20        "img_size"   : 160,21        "normalize"  : False22    },23    "ageClassification": {24        "img_size"   : 160,25        "normalize"  : True,26    },27    "foodClassification":{28        "img_size"   : 224,29        "normalize"  : True,30    },31    "dogClassification": {32        "img_size"   : 224,33        "normalize"  : True,34    },35    "birdsClassification": {36        "img_size"   : 224,37        "normalize"  : True,38    },39    "wildlifeClassification": {40        "img_size"   : 224,41        "normalize"  : False, 42    },43    "landmarkClassification": {44        "img_size"   : 224,45        "normalize"  : True,46    },47    "flowerClassification": {48        "img_size"   : 224,49        "normalize"  : True,50    },51    "animalClassification": {52        "img_size"   : 160,53        "normalize"  : True,54    },55    "sportsClassification": {56        "img_size"   : 224,57        "normalize"  : True,58    },59}60MODEL_LOCATIONS = {61    "catvsDog"               : "./models_quantized/catVsDogMobilenetv224_quantized.tflite",62    "faceMaskClassification" : "./models_quantized/faceMaskClassificationNasNetModel224_quantized.tflite",63    "genderClassification"   : "./models_quantized/GenderModal_quantized.tflite",64    "emotionClassification"  : "./models_quantized/emotionDetection_quantized.tflite",65    "glassesClassification"  : "./models_quantized/glassesDetection_quantized.tflite",66    "ageClassification"      : "./models_quantized/ageClassificationNasNetModel160_quantized.tflite",67    "foodClassification"     : "./models_quantized/foodClassificationMobilenetv2_quantized.tflite",68    "dogClassification"      : "./models_quantized/dogClassificationMobileNet224(78)_quantized.tflite",69    "birdsClassification"    : "./models_quantized/birdClassificationNasNetModel224_quantized.tflite",70    "wildlifeClassification" : "./models_quantized/oregonWildlife224(92)_quantized.tflite",71    "landmarkClassification" : "./models_quantized/landMarks224_quantized.tflite",72    "flowerClassification"   : "./models_quantized/flower299MobileNet244(76)_quantized.tflite",73    "animalClassification"   : "./models_quantized/animalClassification160_quantized.tflite",74    "sportsClassification"   : "./models_quantized/sportsClassification224_quantized.tflite",75}76LABELS = {77    "catvsDog"               : None,78    "faceMaskClassification" : None,79    "genderClassification"   : None,80    "emotionClassification"  : None,81    "glassesClassification"  : None,82    "ageClassification"      : None,83    "foodClassification"     : food_labels,84    "dogClassification"      : dog_breeds_labels,85    "birdsClassification"    : birds_names_labels,86    "wildlifeClassification" : wildlife_labels,87    "landmarkClassification" : landmark_labels,88    "flowerClassification"   : flowers_name_labels,89    "animalClassification"   : animal_labels,90    "sportsClassification"   : sports_labels,91}92MULTIPLE_CLASSES = {93    "catvsDog"               : False,94    "faceMaskClassification" : False,95    "genderClassification"   : False,96    "emotionClassification"  : False,97    "glassesClassification"  : False,98    "ageClassification"      : False,99    "foodClassification"     : True,100    "dogClassification"      : True,101    "birdsClassification"    : True,102    "wildlifeClassification" : True,103    "landmarkClassification" : True,104    "flowerClassification"   : True,105    "animalClassification"   : True,106    "sportsClassification"   : True,...

Automation Testing Tutorials

Learn to execute automation testing from scratch with LambdaTest Learning Hub. Right from setting up the prerequisites to run your first automation test, to following best practices and diving deeper into advanced test scenarios. LambdaTest Learning Hubs compile a list of step-by-step guides to help you be proficient with different test automation frameworks i.e. Selenium, Cypress, TestNG etc.