Change Log

Fritz follows Semantic Versioning

3.2.2

Add pose smoothing.

3.2.0-beta.2

Internal code restructuring.
Add a new predict method for easier predictions:

import Fritz

let image = FritzVisionImage(...)
let model = FritzVisionPoseModel()

let result = try model.predict(image)

3.2.0-beta.1

Updating to Swift 5.
Fixing Linking issues in Objective-C
Add option to resize Style transfer image output to original size.

3.1.7

Fix model version being incorrectly set when downloading model for first time using fetchModel.

3.1.6

Add option to skip cache on model update check.
Add option to delete all versions of a managed model.

3.1.5

Fix loading managed model from prepackaged model for Fritz Features.

3.1.4

Add functionality to load from prepackaged model without keeping it around at runtime.
Fix Fritz framework to properly export available frameworks.

3.1.3

Make loadModel function public to synchronously load models from OTA.

3.1.2

Adding enum for Fritz Paintings to access models individually.

3.1.1

Added override to force processing using Core ML instead of the Vision framework.

3.1.0

Expose ManagedModel on all VisionModels
Add a public protocol for interacting with all FritzVision models.
Fix SwiftIdentifiedModel download (using shared downloader).

3.0.1

Adding FritzVisionMultiPoseModel to Fritz Framework.

3.0.0 Highlights ✨

Added pose estimation feature. You can use the FritzVisionPoseModel to detect human figures in images and video (Documentation).
Added ability to download FritzVision models over the air (Documentation).
Standardized feature APIs across all FritzVision models (API Reference).
Added ability to download custom models by tags (Documentation).
Added user-configurable metadata to Custom Models. (Documentation).
Simplified dependencies. All feature models are now included in the FritzVision target.
Fixed many broken Objective-C interfaces.

Breaking changes

All VisionModelOptions objects no longer can be configured during initialization. You can configure model options after creating an options object:

-  let modelOptions = FritzVisionLabelModelOptions(threshold: 0.5, numResults: 10)
+  let modelOptions = FritzVisionLabelModelOptions()
+  modelOptions.threshold = 0.5
+  modelOptions.numResults = 10

All models now run with a default imageCropAndScaleOption of .scaleFit. You will have to make the following changes to keep the same behavior:

-  let modelOptions = FritzVisionLabelModelOptions()
+  let modelOptions = FritzVisionLabelModelOptions()
+  modelOptions.imageCropAndScaleOption = .centerCrop

This is especially important for the object detection model as the crop and scale option affects the placement of the bounding boxes.

-  let modelOptions = FritzVisionObjectModelOptions()
+  let modelOptions = FritzVisionObjectModelOptions()
+  modelOptions.imageCropAndScaleOption = .centerCrop

Changed access to label and confidence on FritzVisionObject to simplify access:

- visionObject.label.label
- visionObject.label.confidence
+ visionObject.label
+ visionObject.confidence

3.0.0-beta.7

Fixing Pose Detection result filtering

3.0.0-beta.6

Fixing up pose api

3.0.0-beta.5

Adding Models to FritzVision framework

3.0.0-beta.3

Add getter to access metadata

3.0.0-beta.2

Adding in Model Metadata
Adding option to download models using cellular data
Additional functionality on Style Transfer to load from tags.

3.0.0

did a lot of things will update more later.

2.4.1

Change default model settings to scaleFit.
Add FritzVisionFlexibleStyleTransferModel to use for flexible style transfer models.

2.4.0

Fix OTA Model loading issues
Update people, living room, and outdoor segmentation models to use newer versions of models
Don’t use neural engine for image segmentation models to improve performance

2.3.4

Corrupt Over The Air model downloads no longer triggers a crash.

2.3.3

Update People Segmentation model.

2.3.2

Expose original prediction result in FritzVisionSegmentationResult.

2.3.1

Fix Objective-C annotations.

2.3.0

Updated image segmentation processing. Existing models should now be faster!
Added a higher quality People Segmentation Model.
Added ability to download models over the air without needing to include it in the application bundle.

2.2.2

Fix aspect ratio problem on Style Transfer by using Core ML directly.

2.2.1

Fix bug in FritzVision to add better UIImage support.

2.2.0

Add Public FritzVisionStyleModel you can use to add your own style models
Separate painting styles into their own framework.

2.1.0

Finalizing Fritz Vision Segmentation Model

2.1.0-beta.2

Add outdoor scene
Actually use Fritz to report metrics of segmentation model

2.1.0-beta.1

Add CoreMLHelpers package.
Add People and Living Room Segmentation Models.
Add features to FritzVision to resize images.

2.0.0

Built with Swift 4.2

1.1.3

Add Objective C support for FritzVisionStyleModel

1.1.2

Guard against Gzip returning empty data
Fix Fritz Metrics tracking in Feature SDKs

1.1.1

Add new styles to FritzStyleVision.

1.1.0

Add async prediction functions.
Add FritzStyleVision for live video style transfer.

1.0.0

We're now out of Beta! To add machine learning features to your app create an account.
Updating API Client to be more configurable from server.
Update FritzImageOrientation class to better communicate image orientation options.

1.0.0-beta.22

Uses a single shared SessionManager across all models

1.0.0-beta.21

Move some functionality from FritzManagedModel to FritzCore
Add support for including Fritz-Info.plist
Build tests from all frameworks
Session now optional in SwiftIdentifiedModel

1.0.0-beta.20

Add Fritz tracking to label and object models
Speed up sigmoid calculation by using Accelerate functions

1.0.0-beta.19

Use vision framework for Label model predictions
Add FritzVisionObjectModel for real time object detection.
Add FritzVisionObjectModelOptions and FritzVisionLabelModelOptions to configure model prediction behavior.
Built with Swift 4.1.2.