M-3.1 - Adding LesNet_labels file, adding test for inference and mode…

…l services
Thomasbehan · Jun 3, 2024 · 6dedb8d · 6dedb8d
1 parent de8661b
commit 6dedb8d
Show file tree

Hide file tree

Showing 4 changed files with 190 additions and 1 deletion.
diff --git a/models/LesNet_labels.json b/models/LesNet_labels.json
@@ -0,0 +1,30 @@
+[
+  "acrochordon",
+  "actinic keratosis",
+  "AIMP",
+  "angiofibroma or fibrous papule",
+  "angiokeratoma",
+  "angioma",
+  "atypical melanocytic proliferation",
+  "atypical spitz tumor",
+  "basal cell carcinoma",
+  "benign",
+  "cafe-au-lait macule",
+  "clear cell acanthoma",
+  "dermatofibroma",
+  "lentigo NOS",
+  "lentigo simplex",
+  "lichenoid keratosis",
+  "malignant",
+  "melanoma",
+  "neurofibroma",
+  "nevus",
+  "pigmented benign keratosis",
+  "scar",
+  "sebaceous hyperplasia",
+  "seborrheic keratosis",
+  "solar lentigo",
+  "squamous cell carcinoma",
+  "vascular lesion",
+  "verruca"
+]
diff --git a/skinvestigatorai/config/model.py b/skinvestigatorai/config/model.py
@@ -30,5 +30,5 @@ class ModelConfig(object):
     MAX_AUG_PER_IMAGE = 5000000
     TRAIN_DIR = 'data/train'
     MODEL_TYPE = "KERAS"
-    MODEL_NAME = "LesNetM31.keras"
+    MODEL_NAME = "LesNet.keras"
     LABELS_NAME = "LesNet_labels.json"
diff --git a/tests/test_inference_service.py b/tests/test_inference_service.py
@@ -0,0 +1,90 @@
+from io import BytesIO
+from unittest.mock import MagicMock, patch
+
+import numpy as np
+from pyramid.response import Response
+import pytest
+from PIL import Image
+
+from skinvestigatorai.services.inference import Inference
+from skinvestigatorai.services.model import SVModel
+
+
+@pytest.fixture
+def mock_svmodel():
+    sv_model = MagicMock(SVModel)
+    sv_model.load_model.return_value = (MagicMock(), ["class1", "class2"])
+    sv_model.preprocess_image_for_tflite = lambda x: x
+    return sv_model
+
+
+@pytest.fixture
+def inference(mock_svmodel):
+    with patch('skinvestigatorai.services.model.SVModel', return_value=mock_svmodel):
+        return Inference()
+
+
+def create_mock_image():
+    image = Image.new('RGB', (100, 100))
+    img_byte_arr = BytesIO()
+    image.save(img_byte_arr, format='PNG')
+    img_byte_arr = BytesIO(img_byte_arr.getvalue())
+    return img_byte_arr
+
+
+def test_predict_success(inference):
+    mock_image = create_mock_image()
+
+    inference.model.predict = MagicMock(return_value=np.array([[0.1, 0.9]]))
+
+    result = inference.predict(mock_image)
+
+    assert isinstance(result, dict)
+    assert 'prediction' in result
+    assert 'confidence' in result
+
+
+def test_predict_failure(inference):
+    mock_image = create_mock_image()
+
+    inference.model.predict = MagicMock(return_value=np.array([[0.3, 0.2]]))
+
+    result = inference.predict(mock_image)
+
+    assert isinstance(result, Response)
+    assert result.status_code == 400
+
+
+def test_is_image_similar(inference):
+    mock_image = np.random.rand(100, 100, 3)
+
+    inference.dataset_embedding = np.random.rand(2048)
+    inference._predict_similar = MagicMock(return_value=np.random.rand(2048))
+
+    result = inference.is_image_similar(mock_image, threshold=0.5)
+
+    assert result in [True, False]
+
+
+def test__predict_similar_keras(inference):
+    mock_image = np.random.rand(100, 100, 3)
+    inference.model.predict = MagicMock(return_value=np.random.rand(1, 2048))
+
+    with patch('skinvestigatorai.config.model.ModelConfig.MODEL_TYPE', 'KERAS'):
+        result = inference._predict_similar(mock_image)
+
+    assert result is not None
+
+
+def test__predict_similar_tflite(inference):
+    mock_image = np.random.rand(100, 100, 3)
+    inference.model.get_input_details = MagicMock(return_value=[{'index': 0, 'dtype': np.float32}])
+    inference.model.get_output_details = MagicMock(return_value=[{'index': 1}])
+    inference.model.set_tensor = MagicMock()
+    inference.model.invoke = MagicMock()
+    inference.model.get_tensor = MagicMock(return_value=np.random.rand(1, 2048))
+
+    with patch('skinvestigatorai.config.model.ModelConfig.MODEL_TYPE', 'TFLITE'):
+        result = inference._predict_similar(mock_image)
+
+    assert result is not None
diff --git a/tests/test_model_service.py b/tests/test_model_service.py
@@ -0,0 +1,69 @@
+from unittest.mock import patch, MagicMock
+
+import numpy as np
+import pytest
+
+from skinvestigatorai.config.model import ModelConfig
+from skinvestigatorai.services.model import SVModel
+
+
+@pytest.fixture
+def sv_model():
+    return SVModel()
+
+
+def test_create_feature_extractor_tflite(sv_model):
+    sv_model.model_type = 'TFLITE'
+    mock_model = MagicMock()
+    sv_model.model = mock_model
+    sv_model.create_feature_extractor()
+    assert sv_model.feature_extractor == mock_model
+
+
+def test_create_feature_extractor_invalid_model_type(sv_model):
+    sv_model.model_type = 'INVALID'
+    with pytest.raises(ValueError, match="Unsupported model type. Please use 'KERAS' or 'TFLITE'."):
+        sv_model.create_feature_extractor()
+
+
+def test_preprocess_image_for_tflite(sv_model):
+    img = np.random.rand(224, 224, 3).astype(np.float32)
+    processed_img = sv_model.preprocess_image_for_tflite(img)
+    assert processed_img.shape == (ModelConfig.IMG_SIZE[0], ModelConfig.IMG_SIZE[1], 3)
+    assert np.max(processed_img) <= 1.0
+    assert np.min(processed_img) >= 0.0
+
+
+def test_evaluate_model(sv_model):
+    sv_model.model = MagicMock()
+    sv_model.model.evaluate.return_value = [0.5, 0.8, 0.7, 0.6]
+    test_datagen = MagicMock()
+    test_loss, test_acc, test_precision, test_recall = sv_model.evaluate_model(test_datagen)
+    assert test_loss == 0.5
+    assert test_acc == 0.8
+    assert test_precision == 0.7
+    assert test_recall == 0.6
+
+
+@patch('tensorflow.summary.create_file_writer')
+def test_run_experiments(mock_create_file_writer, sv_model):
+    sv_model.run_experiments = MagicMock()
+    train_ds = MagicMock()
+    val_ds = MagicMock()
+    sv_model.run_experiments(train_ds, val_ds)
+    sv_model.run_experiments.assert_called_once_with(train_ds, val_ds)
+
+
+def test_save_model(sv_model):
+    sv_model.model = MagicMock()
+    with patch('builtins.open', MagicMock()):
+        with patch('tensorflow.keras.models.Model.save', MagicMock()):
+            sv_model.save_model()
+            sv_model.model.save.assert_called_once()
+
+
+def test_load_model(sv_model):
+    with patch('os.path.exists', return_value=True):
+        with patch('tensorflow.keras.models.load_model', return_value=MagicMock()):
+            sv_model.load_model()
+            assert isinstance(sv_model.model, MagicMock)