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.
75 lines
2.4 KiB
Bash
75 lines
2.4 KiB
Bash
#!/bin/bash
|
|
|
|
# Copyright (C) 2024 Intel Corporation
|
|
# SPDX-License-Identifier: Apache-2.0
|
|
|
|
# Advanced Anomalib Pipeline Configuration
|
|
# -------------------------------------
|
|
# This example shows how to configure advanced pipeline settings using the CLI.
|
|
|
|
# 1. Training with Custom Components
|
|
# Configure pre-processing, metrics, and visualization
|
|
echo "Training with custom pipeline components..."
|
|
anomalib train \
|
|
--model patchcore \
|
|
--data MVTecAD \
|
|
--data.category bottle \
|
|
--model.backbone resnet18 \
|
|
--model.layers layer2 layer3 \
|
|
--pre_processor.transform.name Compose \
|
|
--pre_processor.transform.transforms "[
|
|
{name: Resize, size: [256, 256]},
|
|
{name: ToTensor},
|
|
{name: Normalize, mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225]}
|
|
]" \
|
|
--metrics "[auroc, f1_score]" \
|
|
|
|
# 2. Advanced Training Configuration
|
|
# Configure training behavior and optimization
|
|
echo -e "\nTraining with advanced settings..."
|
|
anomalib train \
|
|
--model patchcore \
|
|
--data MVTecAD \
|
|
--trainer.max_epochs 1 \
|
|
--trainer.accelerator gpu \
|
|
--trainer.devices 1 \
|
|
--trainer.precision 16 \
|
|
--trainer.deterministic true \
|
|
--optimizer.name Adam \
|
|
--optimizer.lr 0.001 \
|
|
--scheduler.name CosineAnnealingLR \
|
|
--scheduler.T_max 100
|
|
|
|
# 3. Export and Deploy
|
|
# Export the trained model and run inference
|
|
echo -e "\nExporting and running inference..."
|
|
# First, export the model
|
|
anomalib export \
|
|
--model patchcore \
|
|
--weights path/to/weights.ckpt \
|
|
--export_mode onnx \
|
|
--output_path exported_models
|
|
|
|
# Then, run inference with the exported model
|
|
anomalib predict \
|
|
--model patchcore \
|
|
--weights exported_models/model.onnx \
|
|
--input path/to/test/images \
|
|
--output results/predictions \
|
|
|
|
# 4. Hyperparameter Search
|
|
# Run multiple training configurations
|
|
echo -e "\nRunning hyperparameter search..."
|
|
for backbone in "resnet18" "wide_resnet50_2"; do
|
|
for layer_combo in "layer2,layer3" "layer1,layer2,layer3"; do
|
|
IFS=',' read -ra layers <<< "$layer_combo"
|
|
echo "Training with backbone: $backbone, layers: ${layers[*]}"
|
|
anomalib train \
|
|
--model patchcore \
|
|
--data MVTecAD \
|
|
--model.backbone "$backbone" \
|
|
--model.layers "${layers[@]}" \
|
|
--trainer.default_root_dir "results/search/${backbone}_${layer_combo}"
|
|
done
|
|
done
|