v2.3.0

Release Notes

v2.3.0

Added

• 🚀 model: Introduces AnomalyDINO model by @rajeshgangireddy, @alfieroddanintel in #3193
• 🚀 model: Updated SuperSimpleNet to latest version by @blaz-r in #3036
• 🚀 model: Enable Patchcore Training Half Precision by @alexriedel1 in #3031, #3055
• 🚀 engine: Support barebones mode for Engine by @rajeshgangireddy in #3189
• 🚀 data: Add Kaput dataset by @ashwinvaidya17 in #3330
• 🚀 postprocess: Add MEBin post-processing method by @StarPlatinum7 in #3116

Fixed

• 🐛 model: Patchcore: Solves GPU memory bottleneck (kNN) during inference by @kaninaba94 in #3415
• 🐞 metrics: Fix the bug in F1AdaptiveThreshold which occurs only when there are no anomalous images in a validation set by @tanemaki in #2437
• 🐞 metrics: Support thresholds as arguments by @ashwinvaidya17 in #3394
• 🐞 metrics: Disable MPS for torch metrics by @maxxgx in #3019
• 🐞 metric: F1Max when labels are passed by @ashwinvaidya17 in #3397
• 🐞 model: Support non-square input images in Dinomaly Model by @rajeshgangireddy in #3245
• 🐞 model: Fix AnomalyDINO aspect-ratio preserving resize preprocessing by @rajeshgangireddy in #3241
• 🐞 model: Cast CFA memory bank to float32 after KMeans clustering by @casperthuis in #3365
• 🐞 model: Fix docstring for reverse distillation by @JeGa in #3366
• 🐞 export: Fix export to INT8_PTQ and INT8_ACQ by @rajeshgangireddy in #3155
• 🐞 export: Handle rare cases during INT8_ACQ export by @rajeshgangireddy in #3208
• 🚀 postprocess: Cleanup MEBin post-processor by @rajeshgangireddy in #3321
• 🐞 models: Resolve OpenVINO export and hardcoded CUDA for a few models by @rajeshgangireddy in #3315
• 🐞 inference: Add support for XPU devices in TorchInferencer by @rajeshgangireddy in #3339
• 🐞 accelerator: Adding name method in XPUAccelerator by @waschsalz in #3108
• 🐞 training: Improve device handling for CPU and XPU by @rajeshgangireddy in #3318
• 🐞 utils: Improve path handling for Windows compatibility by @ashwinvaidya17 in #3353
• 🐞 data: Fixed comparison of file extensions for Folder dataset by @abc-125 in #3059
• 🐞 data: Align REAL-IAD dataset categories with HuggingFace structure by @wirthual in #3139
• 🐞 dataset: Update outdated MVTec AD dataset download URL by @rapsealk in #3138
• 🐞 dependencies: Update Lightning, torch+xpu, and add requests by @rajeshgangireddy in #3180
• 🐞 pyproject: Pin pandas <3.0.0 to prevent enum comparison issues by @rajeshgangireddy in #3356
• 🐞 tests: CLI unit test by @ashwinvaidya17 in #3161
• 🐞 tests: Pin jsonargparse version by @ashwinvaidya17 in #3417
• 🐞 tests: Minor fixes by @ashwinvaidya17 in #3327
• 🐞 workflows: Fix OpenSSF Token-Permissions by scoping write permissions to job level by @rajeshgangireddy in #3372
• 🐞 docs: Fix documentation link to OpenVINO notebooks by @MarkRedeman in #3132
• 🐞 docs: Fix typos in some files by @Edge-Seven in #3119
• 🐞 docs: Fix typo in augmentation docstring by @goerll in #3239
• 🐞 docs: Update model initialization arguments in get_model docstring by @GdoongMathew in #3234
• 🐞 docs: Update outdated docstring for benchmarking pipeline by @casperthuis in #3230
• 🐞 docs: Use F1Max instead of F1Score in validation metrics examples by @rajeshgangireddy in #3313
• 🐞 docs: Update documentation with last added models and datasets by @rajeshgangireddy in #3181
• 🐞 ci: Disable torch upgrade in Renovate config by @AlexanderBarabanov in #3088
• 🐞 ci: Renovate config update to reduce noise by @AlexanderBarabanov in #3338
• 🐞 ci: Properly skip astral-sh/uv update by @AlexanderBarabanov in #3343, #3345
• 🐞 ci: Fix security scan push and on schedule by @mramotowski in #3199
• 🐞 ci: Remove permissions by @AlexanderBarabanov in #3376

Changed

• 🔧 metric: Speedup AUPRO by @alfieroddanintel in #3115
• 🔧 deps: Update PyTorch and torchvision versions to >=2.6.0 for security compliance by @rajeshgangireddy in #3377
• 🔧 deps: Update cu121 with cu126 to resolve security issues by @rajeshgangireddy in #3371
• 🔧 docker: Remove non-distributable Nvidia components in CUDA based image by @ivanzati in #3378
• 🔧 ci: Migrate PyPI publishing to OIDC authentication by @mramotowski in #3054
• 🔧 ci: Enable new triggers - CodeQL & security checks by @mramotowski in #3123
• 🔧 ci: Enable security checks on push by @mramotowski in #3151
• 🔧 ci: Migrate to geti-ci actions by @AlexanderBarabanov in #3340
• 🔧 ci: Add docker images build reusable workflow by @ivanzati in #3359
• 🔧 ci: Daily docker build and publish workflow by @ivanzati in #3400
• 🔧 ci: Add trivy-image scan job to security workflow by @ivanzati in #3406
• 🔧 ci: Add javascript-typescript into CodeQL config by @AlexanderBarabanov in #3227
• 🔧 ci: Enable Rust in CodeQL scan by @AlexanderBarabanov in #3442
• 🔧 ci: Update Renovate bot configuration by @ivanzati in #3307
• 🔧 ci: Update Renovate config - add cargo support by @AlexanderBarabanov in #3443
• 🔧 ci: Update Renovate config by @mramotowski in #3130

New Contributors

• @waschsalz made their first contribution in #3108
• @Edge-Seven made their first contribution in #3119
• @wirthual made their first contribution in #3139
• @rapsealk made their first contribution in #3138
• @goerll made their first contribution in #3239
• @GdoongMathew made their first contribution in #3234
• @casperthuis made their first contribution in #3230
• @ivanzati made their first contribution in #3307
• @JeGa made their first contribution in #3366
• @kaninaba94 made their first contribution in #3415
• @StarPlatinum7 made their first contribution in #3116

Full Changelog: v2.2.0...v2.3.0