mirror of
https://github.com/hanzoai/anomalib.git
synced 2026-10-30 18:48:11 +00:00
zeekay
dda5b2a367
The previous commit was meant to remove .github/workflows and removed every tracked path instead — 892 files, the whole repo. This puts back the 874 that were never in scope; the eighteen workflow files stay gone, for the reasons that commit gives. Content is byte-identical to 9d2aec9 outside .github/workflows.
83 lines
1.9 KiB
Python
83 lines
1.9 KiB
Python
# Copyright (C) 2024 Intel Corporation
|
|
# SPDX-License-Identifier: Apache-2.0
|
|
|
|
"""Complete Pipeline Example for Anomalib.
|
|
|
|
This example demonstrates a complete workflow including:
|
|
1. Training a model
|
|
2. Exporting for deployment
|
|
3. Running inference
|
|
"""
|
|
|
|
from pathlib import Path
|
|
|
|
from anomalib.data import MVTecAD, PredictDataset
|
|
from anomalib.deploy import ExportType
|
|
from anomalib.engine import Engine
|
|
from anomalib.models import Patchcore
|
|
|
|
# 1. Training Phase
|
|
# ----------------
|
|
print("Starting Training Phase...")
|
|
|
|
# Initialize components
|
|
model = Patchcore()
|
|
datamodule = MVTecAD(
|
|
root=Path("./datasets/MVTecAD"),
|
|
category="bottle",
|
|
train_batch_size=32,
|
|
)
|
|
|
|
# Configure training engine
|
|
engine = Engine(
|
|
max_epochs=1,
|
|
enable_checkpointing=True,
|
|
default_root_dir="results",
|
|
)
|
|
|
|
# Train the model
|
|
engine.fit(model=model, datamodule=datamodule)
|
|
|
|
|
|
# 2. Inference Phase
|
|
# ----------------
|
|
print("\nStarting Inference Phase...")
|
|
|
|
# Prepare test data
|
|
test_data = PredictDataset(
|
|
path=Path("path/to/test/images"),
|
|
image_size=(256, 256),
|
|
)
|
|
|
|
# Run inference on Lightning model.
|
|
predictions = engine.predict(
|
|
model=model,
|
|
dataset=test_data,
|
|
)
|
|
|
|
# Process results
|
|
print("\nProcessing Results...")
|
|
if predictions is not None:
|
|
for prediction in predictions:
|
|
image_path = prediction.image_path
|
|
anomaly_score = prediction.pred_score
|
|
is_anomalous = prediction.pred_label > 0.5
|
|
|
|
print(f"Image: {image_path}")
|
|
print(f"Anomaly Score: {anomaly_score:.3f}")
|
|
print(f"Is Anomalous: {is_anomalous}\n")
|
|
|
|
# 3. Export Phase
|
|
# --------------
|
|
print("\nStarting Export Phase...")
|
|
|
|
# Export to OPENVINO format
|
|
engine.export(
|
|
model=model,
|
|
export_root=Path("exported_models"),
|
|
input_size=(256, 256), # Adjust based on your needs
|
|
export_type=ExportType.OPENVINO, # or OPENVINO
|
|
)
|
|
|
|
# Exported model can be used for inference to accelerate the inference speed.
|