I'm seeing up to 3x-4x slower results on my first prediction on my MLModel. Seems like once you run the first prediction it's heating the model. Couldn't find any official information about it.
What is the right way to pre-heat my model? Call it on an async thread on app launch with dummy data?
As #Alladinian mentioned, starting from iOS 14 you can use the MLModel.load(contentsOf:...) function.
Here is a usage example for pre-loading a local .mlmodel file:
if let url = Bundle.main.url(forResource: "myModel", withExtension: "mlmodelc") {
let config = MLModelConfiguration()
MLModel.load(contentsOf: url, configuration: config) { [weak self] result in
switch result {
case .success(let model):
print("Model loaded and ready.")
let modelWrapper = MyModelWrapper(model: model)
case .failure(let error):
print("Error loading model: \(error)")
}
}
}
If you want to pre-load an externally fetched model, make sure it's compiled using MLModel.compileModel(at:).
Related
I’m trying to convert escaping closure code to async-await code in a certain file. I’m stuck with implementing the async-await part, specifically whether to still use a do block, with catch and resume, or to not use a do block, and assign the line of code with “try await URLSession.shared.dataTask(with: request)” (that's commented out in File1-GettingSelectedRestaurantBusinessDataUsingAsync-AwaitAndNotUsingCodable.swift in this post below, and meant to be used in the solution attempts to this file) to a variable, then use that variable later similar to how the file File2-GettingRestaurantBusinessDataUsingAsync-AwaitAndUsingCodable.swift does, which is posted further below in this post.
*Note: I used async-await and codable to get the restaurant business data for a certain searched city (thats searched by the user) which is done in a different file (and function). The file I’m having trouble with though is for getting the selected restaurant business’s detail info (name, address, business hours, etc.), and I’m not using codable in this file because some of the data I get when doing this URL request, I get by using NSDictionary; not codable.
How do I correctly implement this async-await concept in my file? The file I’m implementing this in is File1-GettingSelectedRestaurantBusinessDataUsingAsync-AwaitAndNotUsingCodable.swift which is posted further below in this post.
*Update: Where I thought my problem lied when first writing up this question post: At the line of code “URLSession.shared.dataTask(with: request) { (data, response, error) in” in File1-GettingSelectedRestaurantBusinessDataUsingAsync-AwaitAndNotUsingCodable.swift, specifically I thought that I should use the form of code that didn’t use the completionHandler (which is commented as V2 in that file), and whether to use a do block after it, and catch, and resume after the do block.
I’ve posted some attempted solutions so far, which are incomplete, since I’m having the problems mentioned in the first paragraph of this post. I know they don’t work, but this is my thought process so far. These solution attempts are below the code files that I’m working with which are posted further below.
I used the following article for learning more about async-await before attempting to make this current implementation: https://www.avanderlee.com/swift/async-await/.
My code:
Code files that I’m working with:
File that I’m attempting to implement this escaping closure to async-await concept in:
File1-GettingSelectedRestaurantBusinessDataUsingAsync-AwaitAndNotUsingCodable.swift:
import Foundation
import UIKit
import CoreLocation
extension UIViewController {
func retrieveSelectedRestaurantDetailViewInfo(
selected_restaurant_business_ID: String) async throws -> SelectedRestaurantDetailViewInfoNotUsingCodable? {
// MARK: Make API Call
let apiKey = "API key"
/// Create URL
let baseURL =
"https://api.yelp.com/v3/businesses/\(selected_restaurant_business_ID)"
let url = URL(string: baseURL)
/// Creating Request
var request = URLRequest(url: url!)
request.setValue("Bearer \(apiKey)", forHTTPHeaderField: "Authorization")
request.httpMethod = "GET"
///Initialize session and task
//V1: Code for using escaping closure version code.
URLSession.shared.dataTask(with: request) { (data, response, error) in
//This version commented out right now, to show where I'm at with this proboem for clarity. This version is included in the solution attempts; both SoultionAttempt1 and SolutionAttempt2.
if let error = error {
completionHandler(nil, error)
}
//V2: Code for what I think is correct for using async-await version code. Not using the completionHandler here.
// URLSession.shared.dataTask(with: request)
do {
/// Read data as JSON
let json = try JSONSerialization.jsonObject(with: data!, options: [])
/// Main dictionary
guard let responseDictionary = json as? NSDictionary else {return}
//Code for accessing restaraunt detail view info, and assigning it to selectedRestarauntDetailViewInfo view model thats a struct.
var selectedVenue = SelectedRestaurantDetailViewInfoNotUsingCodable()
//Rest of code for accessing restaraunt detail view info, and assigning it to seelctedRestarauntDetailViewInfo view model thats a struct.
selectedVenue.name = responseDictionary.value(forKey: "name") as? String
//*Rest of code for getting business info including address, hours, etc..*
//Commented out for now, because am going with version 2 below.
//V1: Uses escaping closure code version.
// completionHandler(selectedVenue, nil)
//V2: Used for Async/Await code version.
return selectedVenue
} catch {
print("Caught error")
}
}.resume()
}
}
*Update: Below is the new file that shows the code that calls the function with the URLRequest in it in File1-GettingSelectedRestaurantBusinessDataUsingAsync-AwaitAndNotUsingCodable.swift, that uses the async-await concept, as mentioned in a response to a comment in this post:
File0.5-FileWithJustCodeThatCallsTheFunctionForMakingTheURLRequestInFile1.swift:
async let responseSelectedVenueDetailViewInfoNotUsingCodable = try await retrieveSelectedRestaurantDetailViewInfo(selected_restaurant_business_ID: venues.id)
File (file I'm referring to is below; is File2-GettingRestaurantBusinessDataUsingAsync-AwaitAndUsingCodable.swift) that uses async-await for getting the initial restaurant business data, after the user selects a city, which I’m using for reference for making the stated change from the escaping closure code to the async-await code in File1-GettingSelectedRestaurantBusinessDataUsingAsync-AwaitAndNotUsingCodable.swift above:
File2-GettingRestaurantBusinessDataUsingAsync-AwaitAndUsingCodable.swift:
import UIKit
import Foundation
import CoreLocation
class YelpApiSelectedRestaurantDetailViewInfo {
private var apiKey: String
init(apiKey: String) {
self.apiKey = apiKey
}
func searchBusinessDetailViewInfo(selectedRestaurantBusinessID: String) async throws -> SelectedRestaurantDetailViewInfo {
var resultsForTheSelectedRestaurantDetailViewInfo: SelectedRestaurantDetailViewInfo
// MARK: Make URL Request.
var urlComponents = URLComponents(string: "https://api.yelp.com/v3/businesses/\(selectedRestaurantBusinessID)")
guard let url = urlComponents?.url else {
throw URLError(.badURL)
}
var request = URLRequest(url: url)
request.setValue("Bearer \(self.apiKey)", forHTTPHeaderField: "Authorization")
let (data, _) = try await URLSession.shared.data(for: request)
let businessDetailViewInfoResults = try JSONDecoder().decode(SelectedRestaurantDetailViewInfo.self, from: data)
resultsForTheSelectedRestaurantDetailViewInfo = businessDetailViewInfoResults
return resultsForTheSelectedRestaurantDetailViewInfo
}
}
Solution Attempts:
*Note:
-Both of the below solution attempt code snippets start at the line of code: URLSession.shared.dataTask(with: request) { (data, response, error) in in File1-GettingSelectedRestaurantBusinessDataUsingAsync-AwaitAndNotUsingCodable.swift, and goes through the rest of that file, and ends at the same end point as that file (as file File1-GettingSelectedRestaurantBusinessDataUsingAsync-AwaitAndNotUsingCodable.swift).
-Also, the file that these solution attempts are to be implemented in is File1-GettingSelectedRestaurantBusinessDataUsingAsync-AwaitAndNotUsingCodable.swift.
SolutionAttempt1-DoesntUseDoBlock.swift:
///Initialize session and task
try await URLSession.shared.dataTask(with: request)
/// Read data as JSON
let json = try JSONSerialization.jsonObject(with: data!, options: [])
/// Main dictionary
guard let responseDictionary = json as? NSDictionary else {return}
//Code for accessing restaraunt detail view info, and assigning it to selectedRestarauntDetailViewInfo view model thats a struct.
var selectedVenue = SelectedRestaurantDetailViewInfoNotUsingCodable()
//Rest of code for accessing restaraunt detail view info, and assigning it to seelctedRestarauntDetailViewInfo view model thats a struct.
selectedVenue.name = responseDictionary.value(forKey: "name") as? String
//*Rest of code for getting business info including address, business hours, etc..*
return selectedVenue
} catch {
print("Caught error")
}
}.resume()
}
}
*Note about the following solution attempt file: The solution attempt here in my opinion (SolutionAttempt2-DoesUseDoBlock.swift) may not have to include indentation for the do block, where the do block is within the scope of the “try await URLSession.shared.dataTask(with: request)” line of code, but I included the below solution attempt to have this indentation, as it would seem that the do block would need to be within the scope of the “try await URLSession.shared.dataTask(with: request)” line of code, and the original file version of File1-GettingSelectedRestaurantBusinessDataUsingAsync-AwaitAndNotUsingCodable.swift, that uses the escaping closure (not the file being edited/worked on here) had the do block within the “URLSession.shared.dataTask(with: request) { (data, response, error) in” line of code’s scope, which is at the same position as the “try await URLSession.shared.dataTask(with: request)” line of code in this SolutionAttempt2-DoesUseDoBlock.swift file below.
SolutionAttempt2-DoesUseDoBlock.swift:
///Initialize session and task
try await URLSession.shared.dataTask(with: request)
do {
/// Read data as JSON
let json = try JSONSerialization.jsonObject(with: data!, options: [])
/// Main dictionary
guard let responseDictionary = json as? NSDictionary else {return}
//Code for accessing restaraunt detail view info, and assigning it to seelctedRestarauntDetailViewInfo view model thats a struct.
var selectedVenue = SelectedRestaurantDetailViewInfoNotUsingCodable()
//Rest of code for accessing restaraunt detail view info, and assigning it to seelctedRestarauntDetailViewInfo view model thats a struct.
selectedVenue.name = responseDictionary.value(forKey: "name") as? String
//*Rest of code for getting business info including address, business hours, etc.*
return selectedVenue
} catch {
print("Caught error")
}
}.resume()
}
}
Thanks!
I am using Firebase Firestore with Swift and SwiftUI. I have two collections Users and Tweets. Each Tweet store the userId of the user who tweeted. I want to fetch all the tweets along with the user information associated with tweet. Since both are stored in different collections at the root level, this poses a problem. Here is one of the solutions.
private func populateUserInfoIntoTweets(tweets: [Tweet]) {
var results = [Tweet]()
tweets.forEach { tweet in
var tweetCopy = tweet
self.db.collection("users").document(tweetCopy.userId).getDocument(as: UserInfo.self) { result in
switch result {
case .success(let userInfo):
tweetCopy.userInfo = userInfo
results.append(tweetCopy)
if results.count == tweets.count {
DispatchQueue.main.async {
self.tweets = results.sorted(by: { t1, t2 in
t1.dateUpdated > t2.dateUpdated
})
}
}
case .failure(let error):
print(error.localizedDescription)
}
}
}
}
Kinda messy!
Another way is to store the documentRef of the User doc instead of the userId.
If I store the User document reference then I am using the following code:
init() {
db.collection("tweets").order(by: "dateUpdated", descending: true)
.addSnapshotListener { snapshot, error in
if let error {
print(error.localizedDescription)
return
}
snapshot?.documents.compactMap { document in
try? document.data(as: Tweet.self)
}.forEach { tweet in
var tweetCopy = tweet
tweetCopy.userDocRef?.getDocument(as: UserInfo.self, completion: { result in
switch result {
case .success(let userInfo):
tweetCopy.userInfo = userInfo
self.tweets.append(tweetCopy)
case .failure(let error):
print(error.localizedDescription)
}
})
}
}
}
The above code does not work as expected. I am wondering there must be a better way. I don't want to store all the user information with the Tweet because it will be storing a lot of extra data.
Recommendations?
The common alternative is to duplicate the necessary information of the user in their tweets, a process known as denormalization. If you come from a background in relational databases this may sounds like blasphemy, but it's actually quite common in NoSQL data modeling (and one of the reasons these databases scale to such massive numbers of readers).
If you want to learn more about such data modeling considerations, I recommend checking out NoSQL data modeling and watching Todd's excellent Get to know Cloud Firestore series.
If you're wondering how to keep the duplicated data up to date, this is probably also a good read: How to write denormalized data in Firebase
I have a video sharing app, and when you save a video to firebase storage it works perfectly for videos that are roughly 1 minute or shorter.
The problem that I am having, is when I try to post a longer video (1 min or greater) it never saves to firebase.
The only thing that I can think of is this error that I am getting, and this error only shows up about 30 seconds after I click the save button:
[BackgroundTask] Background Task 101 ("GTMSessionFetcher-firebasestorage.googleapis.com"), was created over 30 seconds ago. In applications running in the background, this creates a risk of termination. Remember to call UIApplication.endBackgroundTask(_:) for your task in a timely manner to avoid this.
Here is my code to save the video to firebase.
func saveMovie(path: String, file: String, url: URL) {
var backgroundTaskID: UIBackgroundTaskIdentifier?
// Perform the task on a background queue.
DispatchQueue.global().async {
// Request the task asseration and save the ID
backgroundTaskID = UIApplication.shared.beginBackgroundTask(withName: "Finish doing this task", expirationHandler: {
// End the task if time expires
UIApplication.shared.endBackgroundTask(backgroundTaskID!)
backgroundTaskID = UIBackgroundTaskIdentifier.invalid
})
// Send the data synchronously
do {
let movieData = try Data(contentsOf: url)
self.storage.child(path).child("\(file).m4v").putData(movieData)
} catch let error {
fatalError("Error saving movie in saveMovie func. \(error.localizedDescription)")
}
//End the task assertion
UIApplication.shared.endBackgroundTask(backgroundTaskID!)
backgroundTaskID = UIBackgroundTaskIdentifier.invalid
}
}
Any suggestions on how I can allow my video time to upload?
Finally figured this out after a long time...
All you have to do is use .putFile("FileURL") instead of .putdata("Data"). Firebase documentation says you should use putFile() instead of putData() when uploading large files.
But the hard part is for some reason you can't directly upload the movie URL that you get from the didFinishPickingMediaWithInfo function and firebase will just give you an error. So what I did instead was get the data of the movie, save the movie data to a path in the file manager, and use the file manager path URL to upload directly to firebase which worked for me.
//Save movie to Firestore
do {
// Convert movie to Data.
let movieData = try Data(contentsOf: movie)
// Get path so we can save movieData into fileManager and upload to firebase because movie URL does not work, but fileManager url does work.
guard let path = FileManager.default.urls(for: .documentDirectory, in: .userDomainMask).first?.appendingPathComponent(postId!) else { print("Error saving to file manager in addPost func"); return }
do {
try movieData.write(to: path)
// Save the file manager url file to firebase storage
Storage.storage().reference().child("Videos").child("\(postId!).m4v").putFile(from: path, metadata: nil) { metadata, error in
if let error = error {
print("There was an error \(error.localizedDescription)")
} else {
print("Video successfully uploaded.")
}
// Delete video from filemanager because it would take up too much space to save all videos to file manager.
do {
try FileManager.default.removeItem(atPath: path.path)
} catch let error {
print("Error deleting from file manager in addPost func \(error.localizedDescription)")
}
}
} catch let error {
print("Error writing movieData to firebase \(error.localizedDescription)")
}
} catch let error {
print("There was an error adding video in addPost func \(error.localizedDescription)")
}
I am using Create ML to create a .mlmodel file using a Image Classifier project.
Create ML, creates this file for 2 classes.
After dragging the created ML Model file , and adding it to compile sources, below code is able to find the file and create the model.
But the meta data does not show classes, any reason why? How would I know the classes in the mlmodel file?
let error: NSError! = nil
guard let modelURL = Bundle.main.url(forResource: "ObjectDetector", withExtension: "mlmodelc") else {
return
}
do {
let model = try MLModel(contentsOf: modelURL)
let visionModel = try VNCoreMLModel(for: model)
let metaData = model.modelDescription.metadata[.creatorDefinedKey] as! [String:String]
// ERROR: It could not find metaData["classes"]
let allClasses = metaData["classes"]!.components(separatedBy: ",")
let objectRecognition = VNCoreMLRequest(model: visionModel, completionHandler: { (request, error) in
DispatchQueue.main.async(execute: {
// perform all the UI updates on the main queue
if let results = request.results {
self.drawVisionRequestResults(results)
}
})
})
self.requests = [objectRecognition]
} catch let error as NSError {
print("Model loading went wrong: \(error)")
}
Please note I am not getting "Model class has not been generated yet.".
https://stackoverflow.com/questions/462476…
Clicking on ML Model file shows Automatically generated Swift model class.
The class data is part of the mlmodel file but not in the metadata. You can add it to the metadata if you want, but you'll have to write a short Python script for that.
As of iOS 14, there is MLModelDescription.classLabels that also lets you access the list of class names.
I have been trying to have my app stream videos from YouTube, learning from the AppCoda tutorial and so far I have successfully connected to its JSON database, and downloaded the info from the video I need. However, when I feed it to the YouTube player library, it gives me an error:
Error Domain=NSCocoaErrorDomain Code=258 "The file name is invalid."
I checked against the AppCoda example and it seems that my app is working very similarly. I am even printing the videoID before calling the function to make sure that it is correct.
func getVideos(pID: String, completion: (result:String)->()) {
Alamofire.request(.GET, videosURL).validate().responseJSON { response in
switch response.result {
case .Success:
if let value = response.result.value {
let json = JSON(value)
var videoID: [String] = []
for var i=0; i<json["items"].count; i++ {
let newID = json["items"][i]["snippet"]["resourceId"]["videoId"].stringValue
videoID.append(newID)
defaults.setObject(videoID, forKey: "\(pID)videoID")
}
completion(result: "done")
}
case .Failure(let error):
print(error)
completion(result: "done")
}
}
}
print(videoID) //Returns: 7zAxweqsSfo
playerView.loadWithVideoId(videoID)
}
The latest version of the YouTube iOS Helper Library does not include the YTPlayerView-iframe-player.html file needed by the player. See this GitHub issue for related discussion.
The solution is to manually add the YTPlayerView-iframe-player.html to your project, or point Cocoapods at the master branch:
pod 'youtube-ios-player-helper', :git=>'https://github.com/youtube/youtube-ios-player-helper', :commit=>'head'