I am currently working on a 2D pose estimator. I developed a PyTorch vision transformer based model with 17 joints in COCO format for the same and then converted it to CoreML using CoreML tools version 6.2. The model was trained on a custom dataset. However, upon running the converted model on iOS, I observed a significant drop in accuracy. You can see it in this video (https://youtu.be/EfGFrOZQGtU) that demonstrates the outputs of the PyTorch model (on the left) and the CoreML model (on the right). Could you please confirm if this drop in accuracy is expected and suggest any possible solutions to address this issue? Please note that all preprocessing and post-processing techniques remain consistent between the models. P.S. While converting I also got the following warning. : TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs! if x.numel() == 0 and obsolete_torch_version(TORCH_VERSION, (1, 4)): P.P.S. When we initialize the CoreML model on iOS 17.0, we get this error: Validation failure: Invalid Pool kernel width (13), must be [1-8] or 20. Validation failure: Invalid Pool kernel width (9), must be [1-8] or 20. Validation failure: Invalid Pool kernel width (13), must be [1-8] or 20. Validation failure: Invalid Pool kernel width (9), must be [1-8] or 20. Validation failure: Invalid Pool kernel width (13), must be [1-8] or 20. This neural network model does not have a parameter for requested key 'precisionRecallCurves'. Note: only updatable neural network models can provide parameter values and these values are only accessible in the context of an MLUpdateTask completion or progress handler.

I made a model using pytorch and then converted it into a mlmodel file. Next I tried and downloaded (https://developer.apple.com/documentation/vision/recognizing_objects_in_live_capture) which worked! But when I changed the model to my model that I made, the camera worked, but no predictions where shown please h elp!

Hi, this is the 3rd time I'm trying to post this on the forum, apple moderators ignoring it. I'm a deep learning expert with a specialization of image processing. I want to know why I have hundreds of AI models on my Mac that are indexing everything on my computer while it is idle, using programs like neuralhash that I can't find any information about. I can understand if they are being used to enhance the user experience on Spotlight, Siri, Photos, and other applications, but I couldn't find the necessary information on the web. Usually, (spyware) software like this uses them to classify files in an X/Y coordinate system. This feels like a more advanced version of stuxnet. find / -type f -name "*.weights" > ai_models.txt find / -type f -name "*labels*.txt" > ai_model_labels.txt Some of the classes from the files; file_name: SCL_v0.3.1_9c7zcipfrc_558001-labels-v3.txt document_boarding_pass document_check_or_checkbook document_currency_or_bill document_driving_license document_office_badge document_passport document_receipt document_social_security_number hier_curation hier_document hier_negative curation_meme file_name: SceneNet5_detection_labels-v8d.txt CVML_UNKNOWN_999999 aircraft automobile bicycle bird bottle bus canine consumer_electronics feline fruit furniture headgear kite fish computer_monitor motorcycle musical_instrument document people food sign watersport train ungulates watercraft flower appliance sports_equipment tool

I'm trying to create an app that uses artificial intelligence technology. One of the models provided on this website(https://developer.apple.com/machine-learning/models/) will be used. Are there any copyright or legal issues if I create an app using the model provided by this website and distribute it to the App Store?

I'm trying to create an app that uses artificial intelligence technology. One of the models provided on this website(https://developer.apple.com/machine-learning/models/) will be used. Are there any copyright or legal issues if I create an app using the model provided by this website and distribute it to the App Store?

I was watching WWDC20_Model Deployment, but I found that there's no existing documentation backing up this session. Is model deployment dashboard still available in 2024?

Hi, i have been noticing some strange issues with using CoreML models in my app. I am using the Whisper.cpp implementation which has a coreML option. This speeds up the transcribing vs Metal. However every time i use it, the app size inside iphone settings -> General -> Storage increases - specifically the "documents and data" part, the bundle size stays consistent. The Size of the app seems to increase by the same size of the coreml model, and after a few reloads it can increase to over 3-4gb! I thought that maybe the coreml model (which is in the bundle) is being saved to file - but i can't see where, i have tried to use instruments and xcode plus lots of printing out of cache and temp directory etc, deleting the caches etc.. but no effect. I have downloaded the container of the iphone from xcode and inspected it, there are some files stored inthe cache but only a few kbs, and even though the value in the settings-> storage shows a few gb, the container is only a few mb. Please can someone help or give me some guidance on what to do to figure out why the documents and data is increasing? where could this folder be pointing to that is not in the xcode downloaded container?? This is the repo i am using https://github.com/ggerganov/whisper.cpp the swiftui app and objective-C app both do the same thing i am witnessing when using coreml. Thanks in advance for any help, i am totally baffled by this behaviour

for (int i = 0; i < 1000; i++){ double st_tmp = CFAbsoluteTimeGetCurrent(); retBuffer = [self.enhancer enhance:pixelBuffer error:&error]; double et_tmp = CFAbsoluteTimeGetCurrent(); NSLog(@"[enhance once] %f ms ", (et_tmp - st_tmp) * 1000); } When I run a CoreML model using the above code, I notice that the runtime gradually decreases at the beginning. output： [enhance once] 14.965057 ms [enhance once] 12.727022 ms [enhance once] 12.818098 ms [enhance once] 11.829972 ms [enhance once] 11.461020 ms [enhance once] 10.949016 ms [enhance once] 10.712981 ms [enhance once] 10.367990 ms [enhance once] 10.077000 ms [enhance once] 9.699941 ms [enhance once] 9.370089 ms [enhance once] 8.634090 ms [enhance once] 7.659078 ms [enhance once] 7.061005 ms [enhance once] 6.729007 ms [enhance once] 6.603003 ms [enhance once] 6.427050 ms [enhance once] 6.376028 ms [enhance once] 6.509066 ms [enhance once] 6.452084 ms [enhance once] 6.549001 ms [enhance once] 6.616950 ms [enhance once] 6.471038 ms [enhance once] 6.462932 ms [enhance once] 6.443977 ms [enhance once] 6.683946 ms [enhance once] 6.538987 ms [enhance once] 6.628990 ms ... In most deep learning inference frameworks, there is usually a warmup process, but typically, only the first inference is slower. Why does CoreML have a decreasing runtime at the beginning? Is there a way to make only the first inference time longer, while keeping the rest consistent? I use the CoreML model in the (void)display_pixels:(IJKOverlay *)overlay function.

Hi, I try to create some machine learning model for each stock in S&P500 index. When creating the model(Boosted tree model) I try to make it more successfully by doing hyper parameters using GridSearchCV. It takes so long to create one model so I don't want to think of creating all stocks models. I tried to work with CreateML and swift but it looks like it takes longer to run than sklearn on python. My question is how can I make the process faster? is there any hyper parameters on CreateML on swift (I couldn't find it at docs) and how can I run this code on my GPU? (should be much faster).

I know that I can use face detect with CoreML, but I'm wandering that is there any to identify the same person between two images like Photos app.

This model run coreml result is not right, the precision is completely wrong, I posted a PhotoDepthAnythingConv.onnx model: https://github.com/MoonCodeMaster/CoremlErrorModel/tree/main/DepthAnything

Hello, I have created a Neural Network → K Nearest Neighbors Classifier with python. # followed by k-Nearest Neighbors for classification. import coremltools import coremltools.proto.FeatureTypes_pb2 as ft from coremltools.models.nearest_neighbors import KNearestNeighborsClassifierBuilder import copy # Take the SqueezeNet feature extractor from the Turi Create model. base_model = coremltools.models.MLModel("SqueezeNet.mlmodel") base_spec = base_model._spec layers = copy.deepcopy(base_spec.neuralNetworkClassifier.layers) # Delete the softmax and innerProduct layers. The new last layer is # a "flatten" layer that outputs a 1000-element vector. del layers[-1] del layers[-1] preprocessing = base_spec.neuralNetworkClassifier.preprocessing # The Turi Create model is a classifier, which is treated as a special # model type in Core ML. But we need a general-purpose neural network. del base_spec.neuralNetworkClassifier.layers[:] base_spec.neuralNetwork.layers.extend(layers) # Also copy over the image preprocessing options. base_spec.neuralNetwork.preprocessing.extend(preprocessing) # Remove other classifier stuff. base_spec.description.ClearField("metadata") base_spec.description.ClearField("predictedFeatureName") base_spec.description.ClearField("predictedProbabilitiesName") # Remove the old classifier outputs. del base_spec.description.output[:] # Add a new output for the feature vector. output = base_spec.description.output.add() output.name = "features" output.type.multiArrayType.shape.append(1000) output.type.multiArrayType.dataType = ft.ArrayFeatureType.FLOAT32 # Connect the last layer to this new output. base_spec.neuralNetwork.layers[-1].output[0] = "features" # Create the k-NN model. knn_builder = KNearestNeighborsClassifierBuilder(input_name="features", output_name="label", number_of_dimensions=1000, default_class_label="???", number_of_neighbors=3, weighting_scheme="inverse_distance", index_type="linear") knn_spec = knn_builder.spec knn_spec.description.input[0].shortDescription = "Input vector" knn_spec.description.output[0].shortDescription = "Predicted label" knn_spec.description.output[1].shortDescription = "Probabilities for each possible label" knn_builder.set_number_of_neighbors_with_bounds(3, allowed_range=(1, 10)) # Use the same name as in the neural network models, so that we # can use the same code for evaluating both types of model. knn_spec.description.predictedProbabilitiesName = "labelProbability" knn_spec.description.output[1].name = knn_spec.description.predictedProbabilitiesName # Put it all together into a pipeline. pipeline_spec = coremltools.proto.Model_pb2.Model() pipeline_spec.specificationVersion = coremltools._MINIMUM_UPDATABLE_SPEC_VERSION pipeline_spec.isUpdatable = True pipeline_spec.description.input.extend(base_spec.description.input[:]) pipeline_spec.description.output.extend(knn_spec.description.output[:]) pipeline_spec.description.predictedFeatureName = knn_spec.description.predictedFeatureName pipeline_spec.description.predictedProbabilitiesName = knn_spec.description.predictedProbabilitiesName # Add inputs for training. pipeline_spec.description.trainingInput.extend([base_spec.description.input[0]]) pipeline_spec.description.trainingInput[0].shortDescription = "Example image" pipeline_spec.description.trainingInput.extend([knn_spec.description.trainingInput[1]]) pipeline_spec.description.trainingInput[1].shortDescription = "True label" pipeline_spec.pipelineClassifier.pipeline.models.add().CopyFrom(base_spec) pipeline_spec.pipelineClassifier.pipeline.models.add().CopyFrom(knn_spec) pipeline_spec.pipelineClassifier.pipeline.names.extend(["FeatureExtractor", "kNNClassifier"]) coremltools.utils.save_spec(pipeline_spec, "../Models/FaceDetection.mlmodel") it is from the following tutorial: https://machinethink.net/blog/coreml-training-part3/ It Works and I were am to include it into my project: I want to train the model via the MLUpdateTask: ar batchInputs: [MLFeatureProvider] = [] let imageconstraint = (model.model.modelDescription.inputDescriptionsByName["image"]?.imageConstraint) let imageOptions: [MLFeatureValue.ImageOption: Any] = [ .cropAndScale: VNImageCropAndScaleOption.scaleFill.rawValue] var featureProviders = [MLFeatureProvider]() //URLS where images are stored let trainingData = ImageManager.getImagesAndLabel() for data in trainingData{ let label = data.key for imgURL in data.value{ let featureValue = try MLFeatureValue(imageAt: imgURL, constraint: imageconstraint!, options: imageOptions) if let pixelBuffer = featureValue.imageBufferValue{ let featureProvider = FaceDetectionTrainingInput(image: pixelBuffer, label: label) batchInputs.append(featureProvider)}} let trainingData = MLArrayBatchProvider(array: batchInputs) When calling the MLUpdateTask as follows, the context.model from completionHandler is null. Unfortunately there is no other Information available from the compiler. do{ debugPrint(context) try context.model.write(to: ModelManager.targetURL) } catch{ debugPrint("Error saving the model \(error)") } }) updateTask.resume() I get the following error when I want to access the context.model: Thread 5: EXC_BAD_ACCESS (code=1, address=0x0) Can some1 more experienced tell me how to fix this? It seems like I am missing some parameters? I am currently not splitting the Data when training into train and test data. only preprocessing im doing is scaling the image down to 227x227 pixels. Thanks!

Hey, i just created and trained an MLImageClassifier via the MLImageclassifier.train() method (https://developer.apple.com/documentation/createml/mlimageclassifier/train(trainingdata:parameters:sessionparameters:)) For my Trainingdata (MLImageclassifier.DataSource) i am using my directoy structure, so i got an images folder with subfolders of person1, person2, person3 etc. which contain images of the labeled persons (https://developer.apple.com/documentation/createml/mlimageclassifier/datasource/labeleddirectories(at:)) I am saving the checkpoints and sessions in my appdirectory, so i can create an MLIMageClassifier from an exisiting MLSession and/or MLCheckpoint. My question is: is there any way to add new labels, optimally from my directoy strucutre, to an MLImageClassifier which i create from an existing MLCheckpoint/MLSession? So like adding a person4 and training my pretrained Classifier with only that person4. Or is it simply not possible and i have to train from the beginning everytime i want to add a new label? Unfortunately i cannot find anything in the API. Thanks!

I was wondering if there is a quick way to convert a model trained with the open source CRFSuite for use with NLTagger? It seems like retraining should be possible but was wondering if automatic conversion was supported?

Hey, im training an MLImageClassifier via the train()-method: guard let job = try? MLImageClassifier.train(trainingData: trainingData, parameters: modelParameter, sessionParameters: sessionParameters) else{ debugPrint("Training failed") return } Unfortunately the metrics of my MLProgress, which is created from the returning MLJob while training are empty. Code for listening on Progress: job.progress.publisher(for: \.fractionCompleted) .sink{[weak job] fractionCompleted in guard let job = job else { debugPrint("failure in creating job") return } guard let progress = MLProgress(progress: job.progress) else { debugPrint("failure in creating progress") return } print("ProgressPROGRESS: \(progress)") print("Progress: \(fractionCompleted)") } .store(in: &subscriptions) Printing the Progress ends in: MLProgress(elapsedTime: 2.2328420877456665, phase: CreateML.MLPhase.extractingFeatures, itemCount: 32, totalItemCount: Optional(39), metrics: [:]) Got the Same result when listening to MLCheckpoints, Metrics are empty aswell: MLCheckpoint(url: URLPATH.checkpoint, phase: CreateML.MLPhase.extractingFeatures, iteration: 32, date: 2024-04-18 11:21:18 +0000, metrics: [:]) Can some1 tell me how I can access the metrics while training? Thanks!

I'm using Filemaker, with Monkey Bread Software plugin's CoreML features, to find that it can only write to .mlmodelc. Are these (.mlmodel = .mlmodelc) the same? If not, how do you generate a .mlmodelc using XCode. Please let me know, thanks.

I hope this message finds you well. I recently had the opportunity to watch the insightful session titled "Improve Core ML Integration with Async Prediction" and was thoroughly impressed by the depth of information and the practical demonstration provided. The session offered valuable insights that I believe would greatly benefit my ongoing projects and my understanding of Core ML integration. As I am keen on implementing the demonstrated workflows and techniques within my own work, I am reaching out to kindly request access to the source code and any related material presented during the session. Having access to the code would enable me to better understand the concepts discussed and apply them more effectively in real-world scenarios. I believe that being able to review and experiment with the actual code would significantly enhance my learning experience and the implementation efficiency of my projects. It would also serve as a valuable resource for referencing best practices in Core ML integration and async prediction techniques. Thank you very much for considering my request. I greatly appreciate the effort that went into creating such an informative session and am looking forward to potentially exploring the material in greater depth. Best regards, Fabio G.

I have an mlprogram of size 127.2MB it was created using tensorflow and then converted to CoreML. When I request a prediction the amount of memory shoots up to 2-2.5GB every time. I've tried using the optimization techniques in coremltools but nothing seems to work it still shoots up to the same 2-2.5GB of ram every time. I've attached a graph to see it doesn't seem to be a leak as the memory is then going back down.

The CoreML model worked correctly in the “Preview” of “CreateML”. However, after it is put into the Xcode project and replaced the “MobileNetV2” , it did not classify the images correctly, it returned one image with high confidence all the time no matter what image it is . The same code works fine when executed on real device. Can someone please assist on this ?

Hi everyone, is it possible to use a 3D USDZ file to train a model in Create ML, I see there is an image option but it would be good to use these files from Reality Composer from object capture? Or is this in the works for forthcoming Xcode updates? Many Thanks Stuart