Test OpenVINO export

[adrianboguszewski]

📝 Description

@samet-akcay @ashwinvaidya17 is it the right place to test if all models are exportable to OpenVINO?

✨ Changes

Select what type of change your PR is:

• 🐞 Bug fix (non-breaking change which fixes an issue)
• 🔨 Refactor (non-breaking change which refactors the code base)
• 🚀 New feature (non-breaking change which adds functionality)
• 💥 Breaking change (fix or feature that would cause existing functionality to not work as expected)
• 📚 Documentation update
• 🔒 Security update

✅ Checklist

Before you submit your pull request, please make sure you have completed the following steps:

• 📋 I have summarized my changes in the CHANGELOG and followed the guidelines for my type of change (skip for minor changes, documentation updates, and test enhancements).
• 📚 I have made the necessary updates to the documentation (if applicable).
• 🧪 I have written tests that support my changes and prove that my fix is effective or my feature works (if applicable).

[samet-akcay]

Thanks @adrianboguszewski. I have one suggestion to rename framework to export_type. Since framework term is not used in the main code base, the code might flow better if we stick to the export_type term that is used throughout the codebase

[adrianboguszewski]

@samet-akcay your review has been incorporated :)

[nguyenanhtuan1008]

@adrianboguszewski @samet-akcay
sorry, I have a question. I used command line for training: anomalib train --config <path/to/config>. Now I want to export the custom model to TORCH and OpenVINO model, how can I do it?

This my command:
anomalib train --config /workspaces/anomalib/efficient_ad.yaml --data /workspaces/anomalib/mvtec.yaml

mvtec.yaml:

class_path: anomalib.data.MVTec
init_args:
  root: /workspaces/anomalib/datasets/MVTec
  category: mvt_20240209
  image_size: [512, 512]
  center_crop: null
  normalization: imagenet
  train_batch_size: 4
  eval_batch_size: 2
  num_workers: 8
  task: SEGMENTATION
  transform_config_train: null
  transform_config_eval: null
  test_split_mode: FROM_DIR
  test_split_ratio: 0.2
  val_split_mode: SAME_AS_TEST
  val_split_ratio: 0.5
  seed: null

and efficient_ad.yaml:

model:
  class_path: anomalib.models.EfficientAd
  init_args:
    teacher_out_channels: 384
    model_size: M
    lr: 0.0001
    weight_decay: 1.0e-05
    padding: false
    pad_maps: true
    batch_size: 1

metrics:
  pixel:
    - AUROC

trainer:
  max_epochs: 1000
  max_steps: 70000

[adrianboguszewski]

I believe @samet-akcay can answer this question better