diff --git a/aidge_export_cpp/unit_tests/test_examples.py b/aidge_export_cpp/unit_tests/test_examples.py
new file mode 100644
index 0000000000000000000000000000000000000000..1ddf6590acf26b0a4f5e31b43900176551c77a7f
--- /dev/null
+++ b/aidge_export_cpp/unit_tests/test_examples.py
@@ -0,0 +1,46 @@
+import subprocess
+import sys
+import os
+import pytest
+from pathlib import Path
+
+CURRENT_DIR = Path(__file__).parent
+EXAMPLES_DIR = CURRENT_DIR / "../../examples"
+
+# Dictionary of test cases: {id: (script_name, script_args)}
+TEST_CASES = {
+ "lenet-int8-no-args": ("export_LeNet/lenet.py", []),
+ "resnet18-int8-no-args": ("export_ResNet18/resnet18.py", ["--mock_db"])
+}
+
+def generate_test_cases():
+ """Parse TEST_CASES to provide valid pytest params.
+ """
+ for test_id, (script, args) in TEST_CASES.items():
+ yield pytest.param(script, args, id=test_id)
+
+@pytest.mark.parametrize(("script_name", "script_args"), generate_test_cases())
+def test_example_scripts_run_without_error(script_name, script_args):
+ """Basic test to verify that examples script run withoput raising an Error.
+ This test DO NOT check that the examples are working only that they are not broken.
+ """
+ script_path = os.path.join(EXAMPLES_DIR, script_name)
+ result = subprocess.run(
+ [sys.executable, script_path] + script_args, # Or any lightweight args
+ capture_output=True,
+ text=True
+ )
+ assert result.returncode == 0, f"{script_name} failed with error: {result.stderr}"
+
+
+def main():
+ import sys
+
+ print(
+ f"{sys.argv[0]}: Warning: skipped: run with: pytest {sys.argv[0]}",
+ file=sys.stderr,
+)
+
+
+if __name__ == "__main__":
+ main()
diff --git a/examples/export_ResNet18/resnet18.py b/examples/export_ResNet18/resnet18.py
index 07127368eaacb173ad27d3d6f83d0e9605563a03..948a60fa7d56f4624a7941ab696af9e82102f54d 100644
--- a/examples/export_ResNet18/resnet18.py
+++ b/examples/export_ResNet18/resnet18.py
@@ -16,6 +16,7 @@ from PIL import Image
import requests
from pathlib import Path
import subprocess
+from random import randint
# Aidge Modules
import aidge_core
import aidge_onnx
@@ -63,6 +64,10 @@ parser.add_argument(
"DEBUG < INFO < NOTICE < WARN < ERROR < FATAL."
)
)
+
+parser.add_argument("--mock_db", action="store_true", help="Use a mock database instead of real one (TEST ONLY).")
+
+
parser.add_argument(
"--imagenet_path",
type=str,
@@ -187,6 +192,7 @@ def print_cfg():
print(' DEV_MODE = ', DEV_MODE)
print(' IMAGENET_PATH = ', IMAGENET_PATH)
print(' LABEL_PATH = ', LABEL_PATH)
+ print(' MOCK_DB = ', args.mock_db)
print_cfg()
@@ -194,59 +200,68 @@ torch.manual_seed(RNG_SEED)
random.seed(RNG_SEED)
np.random.seed(RNG_SEED)
-backend = "cuda" if USE_CUDA else "cpu"
-
-image_label_pairs = []
-with open(LABEL_PATH, 'r') as f:
- for line in f:
- parts = line.strip().split()
- if len(parts) == 2:
- image_name, label = parts
- image_label_pairs.append((image_name, int(label)))
-
-np.random.seed(RNG_SEED)
-NB_SELECT = max(NB_TEST, NB_CALIB) # Check that NB_TEST and NB_CALIB are fixed
-selected_pairs = image_label_pairs[:NB_SELECT]
-
-# --------------------------------------------------------------
-# CREATE THE SAMPLES
-# --------------------------------------------------------------
-
-transform_val = transforms.Compose([transforms.Resize(256),
- transforms.CenterCrop(224),
- transforms.ToTensor(),
- transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
- ])
-
-tensors = []
-labels = []
-paths = []
-index = 0
-
-for image_name, label in selected_pairs:
- image_path = os.path.join(IMAGENET_PATH, image_name)
- if os.path.exists(image_path):
- try:
- image = Image.open(image_path)
- if image.mode != 'RGB':
- image = image.convert('RGB')
- tensor = transform_val(image)
- tensors.append(tensor)
- labels.append(label)
- paths.append(image_path)
- except Exception as e:
- print(f"Error with image {image_path}: {e}")
-
backend = "cuda" if USE_CUDA else "cpu"
aidge_tensors = []
-for tensor in tensors:
- array = tensor.numpy()
- array = np.reshape(array, (1, 3, 224, 224))
- array = normalize(array)
- aidge_tensor = aidge_core.Tensor(array)
- aidge_tensor.set_backend(backend)
- aidge_tensor.set_datatype(aidge_core.dtype.float32)
- aidge_tensors.append(aidge_tensor)
+labels = []
+if args.mock_db:
+ for i in range(NB_TEST):
+ aidge_tensor = aidge_core.Tensor(dims=(1, 3, 224, 224))
+ aidge_tensor.set_backend(backend)
+ aidge_tensor.set_datatype(aidge_core.dtype.float32)
+ aidge_core.uniform_filler(aidge_tensor, -1.0, 1.0)
+ aidge_tensors.append(aidge_tensor)
+ labels.append(randint(1, 1000))
+else:
+ image_label_pairs = []
+ with open(LABEL_PATH, 'r') as f:
+ for line in f:
+ parts = line.strip().split()
+ if len(parts) == 2:
+ image_name, label = parts
+ image_label_pairs.append((image_name, int(label)))
+
+ np.random.seed(RNG_SEED)
+ NB_SELECT = max(NB_TEST, NB_CALIB) # Check that NB_TEST and NB_CALIB are fixed
+ selected_pairs = image_label_pairs[:NB_SELECT]
+
+ # --------------------------------------------------------------
+ # CREATE THE SAMPLES
+ # --------------------------------------------------------------
+
+ transform_val = transforms.Compose([transforms.Resize(256),
+ transforms.CenterCrop(224),
+ transforms.ToTensor(),
+ transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
+ ])
+
+ tensors = []
+ labels = []
+ paths = []
+ index = 0
+
+ for image_name, label in selected_pairs:
+ image_path = os.path.join(IMAGENET_PATH, image_name)
+ if os.path.exists(image_path):
+ try:
+ image = Image.open(image_path)
+ if image.mode != 'RGB':
+ image = image.convert('RGB')
+ tensor = transform_val(image)
+ tensors.append(tensor)
+ labels.append(label)
+ paths.append(image_path)
+ except Exception as e:
+ print(f"Error with image {image_path}: {e}")
+
+
+ for tensor in tensors:
+ array = tensor.numpy()
+ array = np.reshape(array, (1, 3, 224, 224))
+ array = normalize(array)
+ aidge_tensor = aidge_core.Tensor(array)
+ aidge_tensor.set_backend(backend)
+ aidge_tensor.set_datatype(aidge_core.dtype.float32)
+ aidge_tensors.append(aidge_tensor)
# --------------------------------------------------------------
@@ -260,7 +275,6 @@ Load the .onnx model and perform some usual graph modifications :
- Expand the metaOperators to perform the desired fusions.
"""
-
# Define the target path and filename
file_url = "https://huggingface.co/EclipseAidge/resnet18/resolve/main/resnet18_imagenet_1k.onnx?download=true"
file_path = Path(MODEL_NAME + "_imagenet_1k.onnx")