A library for benchmarking, developing and deploying deep learning anomaly detection algorithms

---

Key Features • Getting Started • Docs • License

Anomalib is a deep learning library that aims to collect state-of-the-art anomaly detection algorithms for benchmarking on both public and private datasets. Anomalib provides several ready-to-use implementations of anomaly detection algorithms described in the recent literature, as well as a set of tools that facilitate the development and implementation of custom models. The library has a strong focus on image-based anomaly detection, where the goal of the algorithm is to identify anomalous images, or anomalous pixel regions within images in a dataset. Anomalib is constantly updated with new algorithms and training/inference extensions, so keep checking!

• The largest public collection of ready-to-use deep learning anomaly detection algorithms and benchmark datasets.
• PyTorch Lightning based model implementations to reduce boilerplate code and limit the implementation efforts to the bare essentials.
• All models can be exported to OpenVINO Intermediate Representation (IR) for accelerated inference on intel hardware.
• A set of inference tools for quick and easy deployment of the standard or custom anomaly detection models.

Following is a guide on how to get started with anomalib. For more details, look at the Documentation.

For getting started with a Jupyter Notebook, please refer to the Notebooks folder of this repository. Additionally, you can refer to a few created by the community:

You can get started with anomalib by just using pip.

pip install anomalib

It is highly recommended to use virtual environment when installing anomalib. For instance, with anaconda, anomalib could be installed as,

yes | conda create -n anomalib_env python=3.10
conda activate anomalib_env
git clone https://github.com/openvinotoolkit/anomalib.git
cd anomalib
pip install -e .

By default python tools/train.py runs PADIM model on leather category from the MVTec AD (CC BY-NC-SA 4.0) dataset.

python tools/train.py    # Train PADIM on MVTec AD leather

Training a model on a specific dataset and category requires further configuration. Each model has its own configuration file, config.yaml , which contains data, model and training configurable parameters. To train a specific model on a specific dataset and category, the config file is to be provided:

python tools/train.py --config <path/to/model/config.yaml>

For example, to train PADIM you can use

python tools/train.py --config src/anomalib/models/padim/config.yaml

Alternatively, a model name could also be provided as an argument, where the scripts automatically finds the corresponding config file.

python tools/train.py --model padim

where the currently available models are:

• CFA
• CFlow
• DFKDE
• DFM
• DRAEM
• EfficientAd
• FastFlow
• GANomaly
• PADIM
• PatchCore
• Reverse Distillation
• STFPM

The pre-trained backbones come from PyTorch Image Models (timm), which are wrapped by FeatureExtractor.

For more information, please check our documentation or the section about feature extraction in "Getting Started with PyTorch Image Models (timm): A Practitioner’s Guide".

Tips:

• Papers With Code has an interface to easily browse models available in timm: https://paperswithcode.com/lib/timm

• You can also find them with the function timm.list_models("resnet*", pretrained=True)

The backbone can be set in the config file, two examples below.

model:
  name: cflow
  backbone: wide_resnet50_2
  pre_trained: true

It is also possible to train on a custom folder dataset. To do so, data section in config.yaml is to be modified as follows:

dataset:
  name: <name-of-the-dataset>
  format: folder
  path: <path/to/folder/dataset>
  normal_dir: normal # name of the folder containing normal images.
  abnormal_dir: abnormal # name of the folder containing abnormal images.
  normal_test_dir: null # name of the folder containing normal test images.
  task: segmentation # classification or segmentation
  mask: <path/to/mask/annotations> #optional
  extensions: null
  split_ratio: 0.2 # ratio of the normal images that will be used to create a test split
  image_size: 256
  train_batch_size: 32
  test_batch_size: 32
  num_workers: 8
  transform_config:
    train: null
    val: null
  create_validation_set: true
  tiling:
    apply: false
    tile_size: null
    stride: null
    remove_border_count: 0
    use_random_tiling: False
    random_tile_count: 16

Anomalib includes multiple tools, including Lightning, Gradio, and OpenVINO inferencers, for performing inference with a trained model.

The following command can be used to run PyTorch Lightning inference from the command line:

python tools/inference/lightning_inference.py -h

As a quick example:

python tools/inference/lightning_inference.py \
    --config src/anomalib/models/padim/config.yaml \
    --weights results/padim/mvtec/bottle/run/weights/model.ckpt \
    --input datasets/MVTec/bottle/test/broken_large/000.png \
    --output results/padim/mvtec/bottle/images

Example OpenVINO Inference:

python tools/inference/openvino_inference.py \
    --weights results/padim/mvtec/bottle/run/openvino/model.bin \
    --metadata results/padim/mvtec/bottle/run/openvino/metadata.json \
    --input datasets/MVTec/bottle/test/broken_large/000.png \
    --output results/padim/mvtec/bottle/images

Ensure that you provide path to metadata.json if you want the normalization to be applied correctly.

You can also use Gradio Inference to interact with the trained models using a UI. Refer to our guide for more details.

A quick example:

python tools/inference/gradio_inference.py \
        --weights results/padim/mvtec/bottle/run/weights/model.ckpt

It is possible to export your model to ONNX or OpenVINO IR

If you want to export your PyTorch model to an OpenVINO model, ensure that export_mode is set to "openvino" in the respective model config.yaml.

optimization:
  export_mode: "openvino" # options: openvino, onnx

To run hyperparameter optimization, use the following command:

python tools/hpo/sweep.py \
    --model padim --model_config ./path_to_config.yaml \
    --sweep_config tools/hpo/sweep.yaml

For more details refer the HPO Documentation

To gather benchmarking data such as throughput across categories, use the following command:

python tools/benchmarking/benchmark.py \
    --config <relative/absolute path>/<paramfile>.yaml

Refer to the Benchmarking Documentation for more details.

Anomablib is integrated with various libraries for experiment tracking such as Comet, tensorboard, and wandb through pytorch lighting loggers.

Below is an example of how to enable logging for hyper-parameters, metrics, model graphs, and predictions on images in the test data-set

visualization:
  log_images: True # log images to the available loggers (if any)
  mode: full # options: ["full", "simple"]

 logging:
  logger: [comet, tensorboard, wandb]
  log_graph: True

For more information, refer to the Logging Documentation

Note: Set your API Key for Comet.ml via comet_ml.init() in interactive python or simply run export COMET_API_KEY=<Your API Key>

This project showcases an end-to-end training and inference pipeline build on top of Anomalib. It provides a web-based UI for uploading MVTec style datasets and training them on the available Anomalib models. It also has sections for calling inference on individual images as well as listing all the images with their predictions in the database.

You can view the project on Github For more details see the Discussion forum

anomalib supports MVTec AD (CC BY-NC-SA 4.0) and BeanTech (CC-BY-SA) for benchmarking and folder for custom dataset training/inference.

MVTec AD dataset is one of the main benchmarks for anomaly detection, and is released under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC BY-NC-SA 4.0).

Note: These metrics are collected with image size of 256 and seed 42. This common setting is used to make model comparisons fair.

If you use this library and love it, use this to cite it 🤗

@misc{anomalib,
      title={Anomalib: A Deep Learning Library for Anomaly Detection},
      author={Samet Akcay and
              Dick Ameln and
              Ashwin Vaidya and
              Barath Lakshmanan and
              Nilesh Ahuja and
              Utku Genc},
      year={2022},
      eprint={2202.08341},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

For those who would like to contribute to the library, see CONTRIBUTING.md for details.

Thank you to all of the people who have already made a contribution - we appreciate your support!