Fork of anomalib library used to integrate some changes that are required for our experiment manager library to work properly. Changes are mainly related to the ability of using torchscript to export models and some metric additions.
To synchronize changes between the original repository and this repo a second remote must be added as follows:
```bash
git remote add sync https://github.com/openvinotoolkit/anomalib.git
git pull sync main
git branch --set-upstream github sync/development
```
If anomalib is updated to sync back again run
```bash
git pull sync development
git merge github master
```
Current version (1.4.*) is synced with anomalib tag 0.7.0.
To publish on pypi, assuming that you have setup authentication properly run:
```bash
poetry publish --build
```
If you are working behind company proxy run instead:
```bash
POETRY_REPOSITORIES_PYPI_URL="https://upload.pypi.org/legacy/" poetry publish -r pypi --build
```
<div align="center">
<img src="https://raw.githubusercontent.com/openvinotoolkit/anomalib/main/docs/source/images/logos/anomalib-wide-blue.png" width="600px">
**A library for benchmarking, developing and deploying deep learning anomaly detection algorithms**
---
[Key Features](#key-features) •
[Getting Started](#getting-started) •
[Docs](https://openvinotoolkit.github.io/anomalib) •
[License](https://github.com/openvinotoolkit/anomalib/blob/main/LICENSE)
[]()
[]()
[]()
[](https://www.comet.com/site/products/ml-experiment-tracking/?utm_source=anomalib&utm_medium=referral)
[](https://www.codacy.com/gh/openvinotoolkit/anomalib/dashboard?utm_source=github.com&utm_medium=referral&utm_content=openvinotoolkit/anomalib&utm_campaign=Badge_Grade)
[]()
[](https://github.com/openvinotoolkit/anomalib/actions/workflows/nightly.yml)
[](https://github.com/openvinotoolkit/anomalib/actions/workflows/pre_merge.yml)
[](https://codecov.io/gh/openvinotoolkit/anomalib)
[](https://github.com/openvinotoolkit/anomalib/actions/workflows/docs.yml)
[](https://pepy.tech/project/anomalib)
</div>
---
# Introduction
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!
## Key features
- The largest public collection of ready-to-use deep learning anomaly detection algorithms and benchmark datasets.
- [**PyTorch Lightning**](https://www.pytorchlightning.ai/) based model implementations to reduce boilerplate code and limit the implementation efforts to the bare essentials.
- All models can be exported to [**OpenVINO**](https://www.intel.com/content/www/us/en/developer/tools/openvino-toolkit/overview.html) Intermediate Representation (IR) for accelerated inference on intel hardware.
- A set of [inference tools](#inference) for quick and easy deployment of the standard or custom anomaly detection models.
---
# Getting Started
Following is a guide on how to get started with `anomalib`. For more details, look at the [Documentation](https://openvinotoolkit.github.io/anomalib).
## Jupyter Notebooks
For getting started with a Jupyter Notebook, please refer to the [Notebooks](notebooks) folder of this repository. Additionally, you can refer to a few created by the community:
## PyPI Install
You can get started with `anomalib` by just using pip.
```bash
pip install anomalib
```
## Local Install
It is highly recommended to use virtual environment when installing anomalib. For instance, with [anaconda](https://www.anaconda.com/products/individual), `anomalib` could be installed as,
```bash
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 .
```
# Training
By default [`python tools/train.py`](tools/train.py)
runs [PADIM](https://arxiv.org/abs/2011.08785) model on `leather` category from the [MVTec AD](https://www.mvtec.com/company/research/datasets/mvtec-ad) [(CC BY-NC-SA 4.0)](https://creativecommons.org/licenses/by-nc-sa/4.0/) dataset.
```bash
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`](src/anomalib/models/padim/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:
```bash
python tools/train.py --config <path/to/model/config.yaml>
```
For example, to train [PADIM](src/anomalib/models/padim) you can use
```bash
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.
```bash
python tools/train.py --model padim
```
where the currently available models are:
- [CFA](src/anomalib/models/cfa)
- [CFlow](src/anomalib/models/cflow)
- [DFKDE](src/anomalib/models/dfkde)
- [DFM](src/anomalib/models/dfm)
- [DRAEM](src/anomalib/models/draem)
- [EfficientAd](src/anomalib/models/efficient_ad)
- [FastFlow](src/anomalib/models/fastflow)
- [GANomaly](src/anomalib/models/ganomaly)
- [PADIM](src/anomalib/models/padim)
- [PatchCore](src/anomalib/models/patchcore)
- [Reverse Distillation](src/anomalib/models/reverse_distillation)
- [STFPM](src/anomalib/models/stfpm)
## Feature extraction & (pre-trained) backbones
The pre-trained backbones come from [PyTorch Image Models (timm)](https://github.com/rwightman/pytorch-image-models), 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"](https://towardsdatascience.com/getting-started-with-pytorch-image-models-timm-a-practitioners-guide-4e77b4bf9055#b83b:~:text=ready%20to%20train!-,Feature%20Extraction,-timm%20models%20also>).
Tips:
- Papers With Code has an interface to easily browse models available in timm: [https://paperswithcode.com/lib/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.
```yaml
model:
name: cflow
backbone: wide_resnet50_2
pre_trained: true
```
## Custom Dataset
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:
```yaml
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
```
# Inference
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:
```bash
python tools/inference/lightning_inference.py -h
```
As a quick example:
```bash
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:
```bash
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](https://openvinotoolkit.github.io/anomalib/tutorials/inference.html#gradio-inference) for more details.
A quick example:
```bash
python tools/inference/gradio_inference.py \
--weights results/padim/mvtec/bottle/run/weights/model.ckpt
```
## Exporting Model to ONNX or OpenVINO IR
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`.
```yaml
optimization:
export_mode: "openvino" # options: openvino, onnx
```
# Hyperparameter Optimization
To run hyperparameter optimization, use the following command:
```bash
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](https://openvinotoolkit.github.io/anomalib/tutorials/hyperparameter_optimization.html)
# Benchmarking
To gather benchmarking data such as throughput across categories, use the following command:
```bash
python tools/benchmarking/benchmark.py \
--config <relative/absolute path>/<paramfile>.yaml
```
Refer to the [Benchmarking Documentation](https://openvinotoolkit.github.io/anomalib/tutorials/benchmarking.html) for more details.
# Experiment Management
Anomablib is integrated with various libraries for experiment tracking such as Comet, tensorboard, and wandb through [pytorch lighting loggers](https://pytorch-lightning.readthedocs.io/en/stable/extensions/logging.html).
Below is an example of how to enable logging for hyper-parameters, metrics, model graphs, and predictions on images in the test data-set
```yaml
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](https://openvinotoolkit.github.io/anomalib/tutorials/logging.html)
Note: Set your API Key for [Comet.ml](https://www.comet.com/signup?utm_source=anomalib&utm_medium=referral) via `comet_ml.init()` in interactive python or simply run `export COMET_API_KEY=<Your API Key>`
# Community Projects
## 1. Web-based Pipeline for Training and Inference
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](https://github.com/vnk8071/anomaly-detection-in-industry-manufacturing/tree/master/anomalib_contribute)
For more details see the [Discussion forum](https://github.com/openvinotoolkit/anomalib/discussions/733)
# Datasets
`anomalib` supports MVTec AD [(CC BY-NC-SA 4.0)](https://creativecommons.org/licenses/by-nc-sa/4.0/) and BeanTech [(CC-BY-SA)](https://creativecommons.org/licenses/by-sa/4.0/legalcode) for benchmarking and `folder` for custom dataset training/inference.
## [MVTec AD Dataset](https://www.mvtec.com/company/research/datasets/mvtec-ad)
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)](https://creativecommons.org/licenses/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.
## Image-Level AUC
| Model | | Avg | Carpet | Grid | Leather | Tile | Wood | Bottle | Cable | Capsule | Hazelnut | Metal Nut | Pill | Screw | Toothbrush | Transistor | Zipper |
| --------------- | -------------- | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :------: | :-------: | :-------: | :-------: | :--------: | :--------: | :-------: |
| **EfficientAd** | **PDN-S** | **0.982** | 0.982 | **1.000** | 0.997 | **1.000** | 0.986 | **1.000** | 0.952 | 0.950 | 0.952 | 0.979 | **0.987** | 0.960 | 0.997 | 0.999 | **0.994** |
| EfficientAd | PDN-M | 0.975 | 0.972 | 0.998 | **1.000** | 0.999 | 0.984 | 0.991 | 0.945 | 0.957 | 0.948 | 0.989 | 0.926 | **0.975** | **1.000** | 0.965 | 0.971 |
| PatchCore | Wide ResNet-50 | 0.980 | 0.984 | 0.959 | 1.000 | **1.000** | 0.989 | 1.000 | **0.990** | **0.982** | 1.000 | 0.994 | 0.924 | 0.960 | 0.933 | **1.000** | 0.982 |
| PatchCore | ResNet-18 | 0.973 | 0.970 | 0.947 | 1.000 | 0.997 | 0.997 | 1.000 | 0.986 | 0.965 | 1.000 | 0.991 | 0.916 | 0.943 | 0.931 | 0.996 | 0.953 |
| CFlow | Wide ResNet-50 | 0.962 | 0.986 | 0.962 | **1.000** | 0.999 | 0.993 | **1.0** | 0.893 | 0.945 | **1.0** | **0.995** | 0.924 | 0.908 | 0.897 | 0.943 | 0.984 |
| CFA | Wide ResNet-50 | 0.956 | 0.978 | 0.961 | 0.990 | 0.999 | 0.994 | 0.998 | 0.979 | 0.872 | 1.000 | **0.995** | 0.946 | 0.703 | **1.000** | 0.957 | 0.967 |
| CFA | ResNet-18 | 0.930 | 0.953 | 0.947 | 0.999 | 1.000 | **1.000** | 0.991 | 0.947 | 0.858 | 0.995 | 0.932 | 0.887 | 0.625 | 0.994 | 0.895 | 0.919 |
| PaDiM | Wide ResNet-50 | 0.950 | **0.995** | 0.942 | **1.000** | 0.974 | 0.993 | 0.999 | 0.878 | 0.927 | 0.964 | 0.989 | 0.939 | 0.845 | 0.942 | 0.976 | 0.882 |
| PaDiM | ResNet-18 | 0.891 | 0.945 | 0.857 | 0.982 | 0.950 | 0.976 | 0.994 | 0.844 | 0.901 | 0.750 | 0.961 | 0.863 | 0.759 | 0.889 | 0.920 | 0.780 |
| DFM | Wide ResNet-50 | 0.943 | 0.855 | 0.784 | 0.997 | 0.995 | 0.975 | 0.999 | 0.969 | 0.924 | 0.978 | 0.939 | 0.962 | 0.873 | 0.969 | 0.971 | 0.961 |
| DFM | ResNet-18 | 0.936 | 0.817 | 0.736 | 0.993 | 0.966 | 0.977 | 1.000 | 0.956 | 0.944 | 0.994 | 0.922 | 0.961 | 0.89 | 0.969 | 0.939 | 0.969 |
| STFPM | Wide ResNet-50 | 0.876 | 0.957 | 0.977 | 0.981 | 0.976 | 0.939 | 0.987 | 0.878 | 0.732 | 0.995 | 0.973 | 0.652 | 0.825 | 0.500 | 0.875 | 0.899 |
| STFPM | ResNet-18 | 0.893 | 0.954 | **0.982** | 0.989 | 0.949 | 0.961 | 0.979 | 0.838 | 0.759 | 0.999 | 0.956 | 0.705 | 0.835 | **0.997** | 0.853 | 0.645 |
| DFKDE | Wide ResNet-50 | 0.774 | 0.708 | 0.422 | 0.905 | 0.959 | 0.903 | 0.936 | 0.746 | 0.853 | 0.736 | 0.687 | 0.749 | 0.574 | 0.697 | 0.843 | 0.892 |
| DFKDE | ResNet-18 | 0.762 | 0.646 | 0.577 | 0.669 | 0.965 | 0.863 | 0.951 | 0.751 | 0.698 | 0.806 | 0.729 | 0.607 | 0.694 | 0.767 | 0.839 | 0.866 |
| GANomaly | | 0.421 | 0.203 | 0.404 | 0.413 | 0.408 | 0.744 | 0.251 | 0.457 | 0.682 | 0.537 | 0.270 | 0.472 | 0.231 | 0.372 | 0.440 | 0.434 |
## Pixel-Level AUC
| Model | | Avg | Carpet | Grid | Leather | Tile | Wood | Bottle | Cable | Capsule | Hazelnut | Metal Nut | Pill | Screw | Toothbrush | Transistor | Zipper |
| ----------- | ------------------ | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :--------: | :--------: | :-------: |
| **CFA** | **Wide ResNet-50** | **0.983** | 0.980 | 0.954 | 0.989 | **0.985** | **0.974** | **0.989** | **0.988** | **0.989** | 0.985 | **0.992** | **0.988** | 0.979 | **0.991** | 0.977 | **0.990** |
| CFA | ResNet-18 | 0.979 | 0.970 | 0.973 | 0.992 | 0.978 | 0.964 | 0.986 | 0.984 | 0.987 | 0.987 | 0.981 | 0.981 | 0.973 | 0.990 | 0.964 | 0.978 |
| PatchCore | Wide ResNet-50 | 0.980 | 0.988 | 0.968 | 0.991 | 0.961 | 0.934 | 0.984 | **0.988** | 0.988 | 0.987 | 0.989 | 0.980 | **0.989** | 0.988 | **0.981** | 0.983 |
| PatchCore | ResNet-18 | 0.976 | 0.986 | 0.955 | 0.990 | 0.943 | 0.933 | 0.981 | 0.984 | 0.986 | 0.986 | 0.986 | 0.974 | 0.991 | 0.988 | 0.974 | 0.983 |
| CFlow | Wide ResNet-50 | 0.971 | 0.986 | 0.968 | 0.993 | 0.968 | 0.924 | 0.981 | 0.955 | 0.988 | **0.990** | 0.982 | 0.983 | 0.979 | 0.985 | 0.897 | 0.980 |
| PaDiM | Wide ResNet-50 | 0.979 | **0.991** | 0.970 | 0.993 | 0.955 | 0.957 | 0.985 | 0.970 | 0.988 | 0.985 | 0.982 | 0.966 | 0.988 | **0.991** | 0.976 | 0.986 |
| PaDiM | ResNet-18 | 0.968 | 0.984 | 0.918 | **0.994** | 0.934 | 0.947 | 0.983 | 0.965 | 0.984 | 0.978 | 0.970 | 0.957 | 0.978 | 0.988 | 0.968 | 0.979 |
| EfficientAd | PDN-S | 0.960 | 0.963 | 0.937 | 0.976 | 0.907 | 0.868 | 0.983 | 0.983 | 0.980 | 0.976 | 0.978 | 0.986 | 0.985 | 0.962 | 0.956 | 0.961 |
| EfficientAd | PDN-M | 0.957 | 0.948 | 0.937 | 0.976 | 0.906 | 0.867 | 0.976 | 0.986 | 0.957 | 0.977 | 0.984 | 0.978 | 0.986 | 0.964 | 0.947 | 0.960 |
| STFPM | Wide ResNet-50 | 0.903 | 0.987 | **0.989** | 0.980 | 0.966 | 0.956 | 0.966 | 0.913 | 0.956 | 0.974 | 0.961 | 0.946 | 0.988 | 0.178 | 0.807 | 0.980 |
| STFPM | ResNet-18 | 0.951 | 0.986 | 0.988 | 0.991 | 0.946 | 0.949 | 0.971 | 0.898 | 0.962 | 0.981 | 0.942 | 0.878 | 0.983 | 0.983 | 0.838 | 0.972 |
## Image F1 Score
| Model | | Avg | Carpet | Grid | Leather | Tile | Wood | Bottle | Cable | Capsule | Hazelnut | Metal Nut | Pill | Screw | Toothbrush | Transistor | Zipper |
| ------------- | ------------------ | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :--------: | :--------: | :-------: |
| **PatchCore** | **Wide ResNet-50** | **0.976** | 0.971 | 0.974 | **1.000** | **1.000** | 0.967 | **1.000** | 0.968 | **0.982** | **1.000** | 0.984 | 0.940 | 0.943 | 0.938 | **1.000** | **0.979** |
| PatchCore | ResNet-18 | 0.970 | 0.949 | 0.946 | **1.000** | 0.98 | 0.992 | **1.000** | **0.978** | 0.969 | **1.000** | **0.989** | 0.940 | 0.932 | 0.935 | 0.974 | 0.967 |
| EfficientAd | PDN-S | 0.970 | 0.966 | **1.000** | 0.995 | **1.000** | 0.975 | **1.000** | 0.907 | 0.956 | 0.897 | 0.978 | 0.982 | 0.944 | 0.984 | 0.988 | 0.983 |
| EfficientAd | PDN-M | 0.966 | 0.977 | 0.991 | **1.000** | 0.994 | 0.967 | 0.984 | 0.922 | 0.969 | 0.884 | 0.984 | 0.952 | 0.955 | 1.000 | 0.929 | 0.979 |
| CFA | Wide ResNet-50 | 0.962 | 0.961 | 0.957 | 0.995 | 0.994 | 0.983 | 0.984 | 0.962 | 0.946 | **1.000** | 0.984 | **0.952** | 0.855 | **1.000** | 0.907 | 0.975 |
| CFA | ResNet-18 | 0.946 | 0.956 | 0.946 | 0.973 | **1.000** | **1.000** | 0.983 | 0.907 | 0.938 | 0.996 | 0.958 | 0.920 | 0.858 | 0.984 | 0.795 | 0.949 |
| CFlow | Wide ResNet-50 | 0.944 | 0.972 | 0.932 | **1.000** | 0.988 | 0.967 | **1.000** | 0.832 | 0.939 | **1.000** | 0.979 | 0.924 | **0.971** | 0.870 | 0.818 | 0.967 |
| PaDiM | Wide ResNet-50 | 0.951 | **0.989** | 0.930 | **1.000** | 0.960 | 0.983 | 0.992 | 0.856 | **0.982** | 0.937 | 0.978 | 0.946 | 0.895 | 0.952 | 0.914 | 0.947 |
| PaDiM | ResNet-18 | 0.916 | 0.930 | 0.893 | 0.984 | 0.934 | 0.952 | 0.976 | 0.858 | 0.960 | 0.836 | 0.974 | 0.932 | 0.879 | 0.923 | 0.796 | 0.915 |
| DFM | Wide ResNet-50 | 0.950 | 0.915 | 0.870 | 0.995 | 0.988 | 0.960 | 0.992 | 0.939 | 0.965 | 0.971 | 0.942 | 0.956 | 0.906 | 0.966 | 0.914 | 0.971 |
| DFM | ResNet-18 | 0.943 | 0.895 | 0.871 | 0.978 | 0.958 | 0.900 | 1.000 | 0.935 | 0.965 | 0.966 | 0.942 | 0.956 | 0.914 | 0.966 | 0.868 | 0.964 |
| STFPM | Wide ResNet-50 | 0.926 | 0.973 | 0.973 | 0.974 | 0.965 | 0.929 | 0.976 | 0.853 | 0.920 | 0.972 | 0.974 | 0.922 | 0.884 | 0.833 | 0.815 | 0.931 |
| STFPM | ResNet-18 | 0.932 | 0.961 | **0.982** | 0.989 | 0.930 | 0.951 | 0.984 | 0.819 | 0.918 | 0.993 | 0.973 | 0.918 | 0.887 | **0.984** | 0.790 | 0.908 |
| DFKDE | Wide ResNet-50 | 0.875 | 0.907 | 0.844 | 0.905 | 0.945 | 0.914 | 0.946 | 0.790 | 0.914 | 0.817 | 0.894 | 0.922 | 0.855 | 0.845 | 0.722 | 0.910 |
| DFKDE | ResNet-18 | 0.872 | 0.864 | 0.844 | 0.854 | 0.960 | 0.898 | 0.942 | 0.793 | 0.908 | 0.827 | 0.894 | 0.916 | 0.859 | 0.853 | 0.756 | 0.916 |
| GANomaly | | 0.834 | 0.864 | 0.844 | 0.852 | 0.836 | 0.863 | 0.863 | 0.760 | 0.905 | 0.777 | 0.894 | 0.916 | 0.853 | 0.833 | 0.571 | 0.881 |
# Reference
If you use this library and love it, use this to cite it 🤗
```tex
@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}
}
```
# Contributing
For those who would like to contribute to the library, see [CONTRIBUTING.md](CONTRIBUTING.md) for details.
Thank you to all of the people who have already made a contribution - we appreciate your support!
<a href="https://github.com/openvinotoolkit/anomalib/graphs/contributors">
<img src="https://contrib.rocks/image?repo=openvinotoolkit/anomalib" />
</a>
Raw data
{
"_id": null,
"home_page": "https://github.com/orobix/quadra",
"name": "anomalib-orobix",
"maintainer": null,
"docs_url": null,
"requires_python": "<3.11,>=3.9",
"maintainer_email": null,
"keywords": null,
"author": "Intel OpenVINO",
"author_email": "help@openvino.intel.com",
"download_url": "https://files.pythonhosted.org/packages/54/32/e735d835ff89909c03ecacc0850490c63f6eef5c4d7849528d55131d85b5/anomalib_orobix-0.7.0.dev143.tar.gz",
"platform": null,
"description": "Fork of anomalib library used to integrate some changes that are required for our experiment manager library to work properly. Changes are mainly related to the ability of using torchscript to export models and some metric additions.\n\nTo synchronize changes between the original repository and this repo a second remote must be added as follows:\n```bash\ngit remote add sync https://github.com/openvinotoolkit/anomalib.git\ngit pull sync main\ngit branch --set-upstream github sync/development\n```\n\nIf anomalib is updated to sync back again run\n```bash\ngit pull sync development\ngit merge github master\n```\n\nCurrent version (1.4.*) is synced with anomalib tag 0.7.0.\n\nTo publish on pypi, assuming that you have setup authentication properly run:\n```bash\npoetry publish --build\n```\n\nIf you are working behind company proxy run instead:\n```bash\nPOETRY_REPOSITORIES_PYPI_URL=\"https://upload.pypi.org/legacy/\" poetry publish -r pypi --build\n```\n\n\n<div align=\"center\">\n\n<img src=\"https://raw.githubusercontent.com/openvinotoolkit/anomalib/main/docs/source/images/logos/anomalib-wide-blue.png\" width=\"600px\">\n\n**A library for benchmarking, developing and deploying deep learning anomaly detection algorithms**\n\n---\n\n[Key Features](#key-features) \u2022\n[Getting Started](#getting-started) \u2022\n[Docs](https://openvinotoolkit.github.io/anomalib) \u2022\n[License](https://github.com/openvinotoolkit/anomalib/blob/main/LICENSE)\n\n[]()\n[]()\n[]()\n[](https://www.comet.com/site/products/ml-experiment-tracking/?utm_source=anomalib&utm_medium=referral)\n[](https://www.codacy.com/gh/openvinotoolkit/anomalib/dashboard?utm_source=github.com&utm_medium=referral&utm_content=openvinotoolkit/anomalib&utm_campaign=Badge_Grade)\n[]()\n[](https://github.com/openvinotoolkit/anomalib/actions/workflows/nightly.yml)\n[](https://github.com/openvinotoolkit/anomalib/actions/workflows/pre_merge.yml)\n[](https://codecov.io/gh/openvinotoolkit/anomalib)\n[](https://github.com/openvinotoolkit/anomalib/actions/workflows/docs.yml)\n[](https://pepy.tech/project/anomalib)\n\n</div>\n\n---\n\n# Introduction\n\nAnomalib 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!\n\n\n\n## Key features\n\n- The largest public collection of ready-to-use deep learning anomaly detection algorithms and benchmark datasets.\n- [**PyTorch Lightning**](https://www.pytorchlightning.ai/) based model implementations to reduce boilerplate code and limit the implementation efforts to the bare essentials.\n- All models can be exported to [**OpenVINO**](https://www.intel.com/content/www/us/en/developer/tools/openvino-toolkit/overview.html) Intermediate Representation (IR) for accelerated inference on intel hardware.\n- A set of [inference tools](#inference) for quick and easy deployment of the standard or custom anomaly detection models.\n\n---\n\n# Getting Started\n\nFollowing is a guide on how to get started with `anomalib`. For more details, look at the [Documentation](https://openvinotoolkit.github.io/anomalib).\n\n## Jupyter Notebooks\n\nFor getting started with a Jupyter Notebook, please refer to the [Notebooks](notebooks) folder of this repository. Additionally, you can refer to a few created by the community:\n\n## PyPI Install\n\nYou can get started with `anomalib` by just using pip.\n\n```bash\npip install anomalib\n```\n\n## Local Install\n\nIt is highly recommended to use virtual environment when installing anomalib. For instance, with [anaconda](https://www.anaconda.com/products/individual), `anomalib` could be installed as,\n\n```bash\nyes | conda create -n anomalib_env python=3.10\nconda activate anomalib_env\ngit clone https://github.com/openvinotoolkit/anomalib.git\ncd anomalib\npip install -e .\n```\n\n# Training\n\nBy default [`python tools/train.py`](tools/train.py)\nruns [PADIM](https://arxiv.org/abs/2011.08785) model on `leather` category from the [MVTec AD](https://www.mvtec.com/company/research/datasets/mvtec-ad) [(CC BY-NC-SA 4.0)](https://creativecommons.org/licenses/by-nc-sa/4.0/) dataset.\n\n```bash\npython tools/train.py # Train PADIM on MVTec AD leather\n```\n\nTraining a model on a specific dataset and category requires further configuration. Each model has its own configuration\nfile, [`config.yaml`](src/anomalib/models/padim/config.yaml)\n, which contains data, model and training configurable parameters. To train a specific model on a specific dataset and\ncategory, the config file is to be provided:\n\n```bash\npython tools/train.py --config <path/to/model/config.yaml>\n```\n\nFor example, to train [PADIM](src/anomalib/models/padim) you can use\n\n```bash\npython tools/train.py --config src/anomalib/models/padim/config.yaml\n```\n\nAlternatively, a model name could also be provided as an argument, where the scripts automatically finds the corresponding config file.\n\n```bash\npython tools/train.py --model padim\n```\n\nwhere the currently available models are:\n\n- [CFA](src/anomalib/models/cfa)\n- [CFlow](src/anomalib/models/cflow)\n- [DFKDE](src/anomalib/models/dfkde)\n- [DFM](src/anomalib/models/dfm)\n- [DRAEM](src/anomalib/models/draem)\n- [EfficientAd](src/anomalib/models/efficient_ad)\n- [FastFlow](src/anomalib/models/fastflow)\n- [GANomaly](src/anomalib/models/ganomaly)\n- [PADIM](src/anomalib/models/padim)\n- [PatchCore](src/anomalib/models/patchcore)\n- [Reverse Distillation](src/anomalib/models/reverse_distillation)\n- [STFPM](src/anomalib/models/stfpm)\n\n## Feature extraction & (pre-trained) backbones\n\nThe pre-trained backbones come from [PyTorch Image Models (timm)](https://github.com/rwightman/pytorch-image-models), which are wrapped by `FeatureExtractor`.\n\nFor more information, please check our documentation or the [section about feature extraction in \"Getting Started with PyTorch Image Models (timm): A Practitioner\u2019s Guide\"](https://towardsdatascience.com/getting-started-with-pytorch-image-models-timm-a-practitioners-guide-4e77b4bf9055#b83b:~:text=ready%20to%20train!-,Feature%20Extraction,-timm%20models%20also>).\n\nTips:\n\n- Papers With Code has an interface to easily browse models available in timm: [https://paperswithcode.com/lib/timm](https://paperswithcode.com/lib/timm)\n\n- You can also find them with the function `timm.list_models(\"resnet*\", pretrained=True)`\n\nThe backbone can be set in the config file, two examples below.\n\n```yaml\nmodel:\n name: cflow\n backbone: wide_resnet50_2\n pre_trained: true\n```\n\n## Custom Dataset\n\nIt is also possible to train on a custom folder dataset. To do so, `data` section in `config.yaml` is to be modified as follows:\n\n```yaml\ndataset:\n name: <name-of-the-dataset>\n format: folder\n path: <path/to/folder/dataset>\n normal_dir: normal # name of the folder containing normal images.\n abnormal_dir: abnormal # name of the folder containing abnormal images.\n normal_test_dir: null # name of the folder containing normal test images.\n task: segmentation # classification or segmentation\n mask: <path/to/mask/annotations> #optional\n extensions: null\n split_ratio: 0.2 # ratio of the normal images that will be used to create a test split\n image_size: 256\n train_batch_size: 32\n test_batch_size: 32\n num_workers: 8\n transform_config:\n train: null\n val: null\n create_validation_set: true\n tiling:\n apply: false\n tile_size: null\n stride: null\n remove_border_count: 0\n use_random_tiling: False\n random_tile_count: 16\n```\n\n# Inference\n\nAnomalib includes multiple tools, including Lightning, Gradio, and OpenVINO inferencers, for performing inference with a trained model.\n\nThe following command can be used to run PyTorch Lightning inference from the command line:\n\n```bash\npython tools/inference/lightning_inference.py -h\n```\n\nAs a quick example:\n\n```bash\npython tools/inference/lightning_inference.py \\\n --config src/anomalib/models/padim/config.yaml \\\n --weights results/padim/mvtec/bottle/run/weights/model.ckpt \\\n --input datasets/MVTec/bottle/test/broken_large/000.png \\\n --output results/padim/mvtec/bottle/images\n```\n\nExample OpenVINO Inference:\n\n```bash\npython tools/inference/openvino_inference.py \\\n --weights results/padim/mvtec/bottle/run/openvino/model.bin \\\n --metadata results/padim/mvtec/bottle/run/openvino/metadata.json \\\n --input datasets/MVTec/bottle/test/broken_large/000.png \\\n --output results/padim/mvtec/bottle/images\n```\n\n> Ensure that you provide path to `metadata.json` if you want the normalization to be applied correctly.\n\nYou can also use Gradio Inference to interact with the trained models using a UI. Refer to our [guide](https://openvinotoolkit.github.io/anomalib/tutorials/inference.html#gradio-inference) for more details.\n\nA quick example:\n\n```bash\npython tools/inference/gradio_inference.py \\\n --weights results/padim/mvtec/bottle/run/weights/model.ckpt\n```\n\n## Exporting Model to ONNX or OpenVINO IR\n\nIt is possible to export your model to ONNX or OpenVINO IR\n\nIf 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`.\n\n```yaml\noptimization:\n export_mode: \"openvino\" # options: openvino, onnx\n```\n\n# Hyperparameter Optimization\n\nTo run hyperparameter optimization, use the following command:\n\n```bash\npython tools/hpo/sweep.py \\\n --model padim --model_config ./path_to_config.yaml \\\n --sweep_config tools/hpo/sweep.yaml\n```\n\nFor more details refer the [HPO Documentation](https://openvinotoolkit.github.io/anomalib/tutorials/hyperparameter_optimization.html)\n\n# Benchmarking\n\nTo gather benchmarking data such as throughput across categories, use the following command:\n\n```bash\npython tools/benchmarking/benchmark.py \\\n --config <relative/absolute path>/<paramfile>.yaml\n```\n\nRefer to the [Benchmarking Documentation](https://openvinotoolkit.github.io/anomalib/tutorials/benchmarking.html) for more details.\n\n# Experiment Management\n\nAnomablib is integrated with various libraries for experiment tracking such as Comet, tensorboard, and wandb through [pytorch lighting loggers](https://pytorch-lightning.readthedocs.io/en/stable/extensions/logging.html).\n\nBelow is an example of how to enable logging for hyper-parameters, metrics, model graphs, and predictions on images in the test data-set\n\n```yaml\nvisualization:\n log_images: True # log images to the available loggers (if any)\n mode: full # options: [\"full\", \"simple\"]\n\n logging:\n logger: [comet, tensorboard, wandb]\n log_graph: True\n```\n\nFor more information, refer to the [Logging Documentation](https://openvinotoolkit.github.io/anomalib/tutorials/logging.html)\n\nNote: Set your API Key for [Comet.ml](https://www.comet.com/signup?utm_source=anomalib&utm_medium=referral) via `comet_ml.init()` in interactive python or simply run `export COMET_API_KEY=<Your API Key>`\n\n# Community Projects\n\n## 1. Web-based Pipeline for Training and Inference\n\nThis 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.\n\nYou can view the project on [Github](https://github.com/vnk8071/anomaly-detection-in-industry-manufacturing/tree/master/anomalib_contribute)\nFor more details see the [Discussion forum](https://github.com/openvinotoolkit/anomalib/discussions/733)\n\n# Datasets\n\n`anomalib` supports MVTec AD [(CC BY-NC-SA 4.0)](https://creativecommons.org/licenses/by-nc-sa/4.0/) and BeanTech [(CC-BY-SA)](https://creativecommons.org/licenses/by-sa/4.0/legalcode) for benchmarking and `folder` for custom dataset training/inference.\n\n## [MVTec AD Dataset](https://www.mvtec.com/company/research/datasets/mvtec-ad)\n\nMVTec AD dataset is one of the main benchmarks for anomaly detection, and is released under the\nCreative Commons Attribution-NonCommercial-ShareAlike 4.0 International License [(CC BY-NC-SA 4.0)](https://creativecommons.org/licenses/by-nc-sa/4.0/).\n\n> Note: These metrics are collected with image size of 256 and seed `42`. This common setting is used to make model comparisons fair.\n\n## Image-Level AUC\n\n| Model | | Avg | Carpet | Grid | Leather | Tile | Wood | Bottle | Cable | Capsule | Hazelnut | Metal Nut | Pill | Screw | Toothbrush | Transistor | Zipper |\n| --------------- | -------------- | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :------: | :-------: | :-------: | :-------: | :--------: | :--------: | :-------: |\n| **EfficientAd** | **PDN-S** | **0.982** | 0.982 | **1.000** | 0.997 | **1.000** | 0.986 | **1.000** | 0.952 | 0.950 | 0.952 | 0.979 | **0.987** | 0.960 | 0.997 | 0.999 | **0.994** |\n| EfficientAd | PDN-M | 0.975 | 0.972 | 0.998 | **1.000** | 0.999 | 0.984 | 0.991 | 0.945 | 0.957 | 0.948 | 0.989 | 0.926 | **0.975** | **1.000** | 0.965 | 0.971 |\n| PatchCore | Wide ResNet-50 | 0.980 | 0.984 | 0.959 | 1.000 | **1.000** | 0.989 | 1.000 | **0.990** | **0.982** | 1.000 | 0.994 | 0.924 | 0.960 | 0.933 | **1.000** | 0.982 |\n| PatchCore | ResNet-18 | 0.973 | 0.970 | 0.947 | 1.000 | 0.997 | 0.997 | 1.000 | 0.986 | 0.965 | 1.000 | 0.991 | 0.916 | 0.943 | 0.931 | 0.996 | 0.953 |\n| CFlow | Wide ResNet-50 | 0.962 | 0.986 | 0.962 | **1.000** | 0.999 | 0.993 | **1.0** | 0.893 | 0.945 | **1.0** | **0.995** | 0.924 | 0.908 | 0.897 | 0.943 | 0.984 |\n| CFA | Wide ResNet-50 | 0.956 | 0.978 | 0.961 | 0.990 | 0.999 | 0.994 | 0.998 | 0.979 | 0.872 | 1.000 | **0.995** | 0.946 | 0.703 | **1.000** | 0.957 | 0.967 |\n| CFA | ResNet-18 | 0.930 | 0.953 | 0.947 | 0.999 | 1.000 | **1.000** | 0.991 | 0.947 | 0.858 | 0.995 | 0.932 | 0.887 | 0.625 | 0.994 | 0.895 | 0.919 |\n| PaDiM | Wide ResNet-50 | 0.950 | **0.995** | 0.942 | **1.000** | 0.974 | 0.993 | 0.999 | 0.878 | 0.927 | 0.964 | 0.989 | 0.939 | 0.845 | 0.942 | 0.976 | 0.882 |\n| PaDiM | ResNet-18 | 0.891 | 0.945 | 0.857 | 0.982 | 0.950 | 0.976 | 0.994 | 0.844 | 0.901 | 0.750 | 0.961 | 0.863 | 0.759 | 0.889 | 0.920 | 0.780 |\n| DFM | Wide ResNet-50 | 0.943 | 0.855 | 0.784 | 0.997 | 0.995 | 0.975 | 0.999 | 0.969 | 0.924 | 0.978 | 0.939 | 0.962 | 0.873 | 0.969 | 0.971 | 0.961 |\n| DFM | ResNet-18 | 0.936 | 0.817 | 0.736 | 0.993 | 0.966 | 0.977 | 1.000 | 0.956 | 0.944 | 0.994 | 0.922 | 0.961 | 0.89 | 0.969 | 0.939 | 0.969 |\n| STFPM | Wide ResNet-50 | 0.876 | 0.957 | 0.977 | 0.981 | 0.976 | 0.939 | 0.987 | 0.878 | 0.732 | 0.995 | 0.973 | 0.652 | 0.825 | 0.500 | 0.875 | 0.899 |\n| STFPM | ResNet-18 | 0.893 | 0.954 | **0.982** | 0.989 | 0.949 | 0.961 | 0.979 | 0.838 | 0.759 | 0.999 | 0.956 | 0.705 | 0.835 | **0.997** | 0.853 | 0.645 |\n| DFKDE | Wide ResNet-50 | 0.774 | 0.708 | 0.422 | 0.905 | 0.959 | 0.903 | 0.936 | 0.746 | 0.853 | 0.736 | 0.687 | 0.749 | 0.574 | 0.697 | 0.843 | 0.892 |\n| DFKDE | ResNet-18 | 0.762 | 0.646 | 0.577 | 0.669 | 0.965 | 0.863 | 0.951 | 0.751 | 0.698 | 0.806 | 0.729 | 0.607 | 0.694 | 0.767 | 0.839 | 0.866 |\n| GANomaly | | 0.421 | 0.203 | 0.404 | 0.413 | 0.408 | 0.744 | 0.251 | 0.457 | 0.682 | 0.537 | 0.270 | 0.472 | 0.231 | 0.372 | 0.440 | 0.434 |\n\n## Pixel-Level AUC\n\n| Model | | Avg | Carpet | Grid | Leather | Tile | Wood | Bottle | Cable | Capsule | Hazelnut | Metal Nut | Pill | Screw | Toothbrush | Transistor | Zipper |\n| ----------- | ------------------ | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :--------: | :--------: | :-------: |\n| **CFA** | **Wide ResNet-50** | **0.983** | 0.980 | 0.954 | 0.989 | **0.985** | **0.974** | **0.989** | **0.988** | **0.989** | 0.985 | **0.992** | **0.988** | 0.979 | **0.991** | 0.977 | **0.990** |\n| CFA | ResNet-18 | 0.979 | 0.970 | 0.973 | 0.992 | 0.978 | 0.964 | 0.986 | 0.984 | 0.987 | 0.987 | 0.981 | 0.981 | 0.973 | 0.990 | 0.964 | 0.978 |\n| PatchCore | Wide ResNet-50 | 0.980 | 0.988 | 0.968 | 0.991 | 0.961 | 0.934 | 0.984 | **0.988** | 0.988 | 0.987 | 0.989 | 0.980 | **0.989** | 0.988 | **0.981** | 0.983 |\n| PatchCore | ResNet-18 | 0.976 | 0.986 | 0.955 | 0.990 | 0.943 | 0.933 | 0.981 | 0.984 | 0.986 | 0.986 | 0.986 | 0.974 | 0.991 | 0.988 | 0.974 | 0.983 |\n| CFlow | Wide ResNet-50 | 0.971 | 0.986 | 0.968 | 0.993 | 0.968 | 0.924 | 0.981 | 0.955 | 0.988 | **0.990** | 0.982 | 0.983 | 0.979 | 0.985 | 0.897 | 0.980 |\n| PaDiM | Wide ResNet-50 | 0.979 | **0.991** | 0.970 | 0.993 | 0.955 | 0.957 | 0.985 | 0.970 | 0.988 | 0.985 | 0.982 | 0.966 | 0.988 | **0.991** | 0.976 | 0.986 |\n| PaDiM | ResNet-18 | 0.968 | 0.984 | 0.918 | **0.994** | 0.934 | 0.947 | 0.983 | 0.965 | 0.984 | 0.978 | 0.970 | 0.957 | 0.978 | 0.988 | 0.968 | 0.979 |\n| EfficientAd | PDN-S | 0.960 | 0.963 | 0.937 | 0.976 | 0.907 | 0.868 | 0.983 | 0.983 | 0.980 | 0.976 | 0.978 | 0.986 | 0.985 | 0.962 | 0.956 | 0.961 |\n| EfficientAd | PDN-M | 0.957 | 0.948 | 0.937 | 0.976 | 0.906 | 0.867 | 0.976 | 0.986 | 0.957 | 0.977 | 0.984 | 0.978 | 0.986 | 0.964 | 0.947 | 0.960 |\n| STFPM | Wide ResNet-50 | 0.903 | 0.987 | **0.989** | 0.980 | 0.966 | 0.956 | 0.966 | 0.913 | 0.956 | 0.974 | 0.961 | 0.946 | 0.988 | 0.178 | 0.807 | 0.980 |\n| STFPM | ResNet-18 | 0.951 | 0.986 | 0.988 | 0.991 | 0.946 | 0.949 | 0.971 | 0.898 | 0.962 | 0.981 | 0.942 | 0.878 | 0.983 | 0.983 | 0.838 | 0.972 |\n\n## Image F1 Score\n\n| Model | | Avg | Carpet | Grid | Leather | Tile | Wood | Bottle | Cable | Capsule | Hazelnut | Metal Nut | Pill | Screw | Toothbrush | Transistor | Zipper |\n| ------------- | ------------------ | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :--------: | :--------: | :-------: |\n| **PatchCore** | **Wide ResNet-50** | **0.976** | 0.971 | 0.974 | **1.000** | **1.000** | 0.967 | **1.000** | 0.968 | **0.982** | **1.000** | 0.984 | 0.940 | 0.943 | 0.938 | **1.000** | **0.979** |\n| PatchCore | ResNet-18 | 0.970 | 0.949 | 0.946 | **1.000** | 0.98 | 0.992 | **1.000** | **0.978** | 0.969 | **1.000** | **0.989** | 0.940 | 0.932 | 0.935 | 0.974 | 0.967 |\n| EfficientAd | PDN-S | 0.970 | 0.966 | **1.000** | 0.995 | **1.000** | 0.975 | **1.000** | 0.907 | 0.956 | 0.897 | 0.978 | 0.982 | 0.944 | 0.984 | 0.988 | 0.983 |\n| EfficientAd | PDN-M | 0.966 | 0.977 | 0.991 | **1.000** | 0.994 | 0.967 | 0.984 | 0.922 | 0.969 | 0.884 | 0.984 | 0.952 | 0.955 | 1.000 | 0.929 | 0.979 |\n| CFA | Wide ResNet-50 | 0.962 | 0.961 | 0.957 | 0.995 | 0.994 | 0.983 | 0.984 | 0.962 | 0.946 | **1.000** | 0.984 | **0.952** | 0.855 | **1.000** | 0.907 | 0.975 |\n| CFA | ResNet-18 | 0.946 | 0.956 | 0.946 | 0.973 | **1.000** | **1.000** | 0.983 | 0.907 | 0.938 | 0.996 | 0.958 | 0.920 | 0.858 | 0.984 | 0.795 | 0.949 |\n| CFlow | Wide ResNet-50 | 0.944 | 0.972 | 0.932 | **1.000** | 0.988 | 0.967 | **1.000** | 0.832 | 0.939 | **1.000** | 0.979 | 0.924 | **0.971** | 0.870 | 0.818 | 0.967 |\n| PaDiM | Wide ResNet-50 | 0.951 | **0.989** | 0.930 | **1.000** | 0.960 | 0.983 | 0.992 | 0.856 | **0.982** | 0.937 | 0.978 | 0.946 | 0.895 | 0.952 | 0.914 | 0.947 |\n| PaDiM | ResNet-18 | 0.916 | 0.930 | 0.893 | 0.984 | 0.934 | 0.952 | 0.976 | 0.858 | 0.960 | 0.836 | 0.974 | 0.932 | 0.879 | 0.923 | 0.796 | 0.915 |\n| DFM | Wide ResNet-50 | 0.950 | 0.915 | 0.870 | 0.995 | 0.988 | 0.960 | 0.992 | 0.939 | 0.965 | 0.971 | 0.942 | 0.956 | 0.906 | 0.966 | 0.914 | 0.971 |\n| DFM | ResNet-18 | 0.943 | 0.895 | 0.871 | 0.978 | 0.958 | 0.900 | 1.000 | 0.935 | 0.965 | 0.966 | 0.942 | 0.956 | 0.914 | 0.966 | 0.868 | 0.964 |\n| STFPM | Wide ResNet-50 | 0.926 | 0.973 | 0.973 | 0.974 | 0.965 | 0.929 | 0.976 | 0.853 | 0.920 | 0.972 | 0.974 | 0.922 | 0.884 | 0.833 | 0.815 | 0.931 |\n| STFPM | ResNet-18 | 0.932 | 0.961 | **0.982** | 0.989 | 0.930 | 0.951 | 0.984 | 0.819 | 0.918 | 0.993 | 0.973 | 0.918 | 0.887 | **0.984** | 0.790 | 0.908 |\n| DFKDE | Wide ResNet-50 | 0.875 | 0.907 | 0.844 | 0.905 | 0.945 | 0.914 | 0.946 | 0.790 | 0.914 | 0.817 | 0.894 | 0.922 | 0.855 | 0.845 | 0.722 | 0.910 |\n| DFKDE | ResNet-18 | 0.872 | 0.864 | 0.844 | 0.854 | 0.960 | 0.898 | 0.942 | 0.793 | 0.908 | 0.827 | 0.894 | 0.916 | 0.859 | 0.853 | 0.756 | 0.916 |\n| GANomaly | | 0.834 | 0.864 | 0.844 | 0.852 | 0.836 | 0.863 | 0.863 | 0.760 | 0.905 | 0.777 | 0.894 | 0.916 | 0.853 | 0.833 | 0.571 | 0.881 |\n\n# Reference\n\nIf you use this library and love it, use this to cite it \ud83e\udd17\n\n```tex\n@misc{anomalib,\n title={Anomalib: A Deep Learning Library for Anomaly Detection},\n author={Samet Akcay and\n Dick Ameln and\n Ashwin Vaidya and\n Barath Lakshmanan and\n Nilesh Ahuja and\n Utku Genc},\n year={2022},\n eprint={2202.08341},\n archivePrefix={arXiv},\n primaryClass={cs.CV}\n}\n```\n\n# Contributing\n\nFor those who would like to contribute to the library, see [CONTRIBUTING.md](CONTRIBUTING.md) for details.\n\nThank you to all of the people who have already made a contribution - we appreciate your support!\n\n<a href=\"https://github.com/openvinotoolkit/anomalib/graphs/contributors\">\n <img src=\"https://contrib.rocks/image?repo=openvinotoolkit/anomalib\" />\n</a>\n",
"bugtrack_url": null,
"license": "Apache-2.0",
"summary": "Orobix anomalib fork",
"version": "0.7.0.dev143",
"project_urls": {
"Homepage": "https://github.com/orobix/quadra",
"Repository": "https://github.com/orobix/quadra"
},
"split_keywords": [],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "e3c09d78a83369654de7bcf6f11831510a77ca4ea84e8f7ce5de01c3e876835f",
"md5": "c57c32df5679607ddc35cabcfa8cf4be",
"sha256": "5d2d3d4e5cc5efb4249de3edb8f7d117c4b3cae919bb1eb5d403517a013cb3a9"
},
"downloads": -1,
"filename": "anomalib_orobix-0.7.0.dev143-py3-none-any.whl",
"has_sig": false,
"md5_digest": "c57c32df5679607ddc35cabcfa8cf4be",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": "<3.11,>=3.9",
"size": 395527,
"upload_time": "2024-10-24T15:04:22",
"upload_time_iso_8601": "2024-10-24T15:04:22.824658Z",
"url": "https://files.pythonhosted.org/packages/e3/c0/9d78a83369654de7bcf6f11831510a77ca4ea84e8f7ce5de01c3e876835f/anomalib_orobix-0.7.0.dev143-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "5432e735d835ff89909c03ecacc0850490c63f6eef5c4d7849528d55131d85b5",
"md5": "9dfc9aef5cb6fc30678059259fabab2e",
"sha256": "d3f3a9f8e451a7430ccdacaaea5ee7d2720cc9bd1365c37287c75112dbf0959e"
},
"downloads": -1,
"filename": "anomalib_orobix-0.7.0.dev143.tar.gz",
"has_sig": false,
"md5_digest": "9dfc9aef5cb6fc30678059259fabab2e",
"packagetype": "sdist",
"python_version": "source",
"requires_python": "<3.11,>=3.9",
"size": 275279,
"upload_time": "2024-10-24T15:04:25",
"upload_time_iso_8601": "2024-10-24T15:04:25.088666Z",
"url": "https://files.pythonhosted.org/packages/54/32/e735d835ff89909c03ecacc0850490c63f6eef5c4d7849528d55131d85b5/anomalib_orobix-0.7.0.dev143.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-10-24 15:04:25",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "orobix",
"github_project": "quadra",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "anomalib-orobix"
}
JSON
