DEV Community: Sadman Samee

Intelligence in Mobile Applications.

Sadman Samee — Thu, 12 Sep 2019 18:12:33 +0000

Machine learning is a cool topic nowadays, people are excited about Google Assistant, Apple Siri, Microsoft Cortana, Amazon Alexa, self-driving cars, robots, even apps like Uber, Snapchat are using Machine Learning to enhance the experience and of course for smart pieces of stuff.
If you are interested in building kick-ass apps with intelligence that includes features such as speech recognition, face detection, natural language processing, prediction or anything like that, it's not always an easy option to develop a machine learning system from the ground for any indie developers or small to medium or even for a large team because of it's cost, time (coffee) consumption, therefore, there are many open-source machine learning platforms like Tensorflow Lite, Pytorch, Core ML, services like Dialogflow, Wit.ai, Google Cloud Machine Learning APIs, Microsoft Cognitive Services, IBM Watson are available to the developers to get things done in a very convenient way.
Here we will explore various machine learning libraries, APIs, and services.

Tensorflow Lite

TensorFlow™ is an open-source software library for numerical computation using data flow graphs. Nodes in the graph represent mathematical operations, while the graph edges represent the multidimensional data arrays (tensors) communicated between them and TensorFlow Lite is TensorFlow's lightweight solution for mobile and embedded devices. It enables on-device machine learning inference with low latency and small binary size.
Android app with Tensorflow.You can build intelligent apps using Tensorflow Lite and even it will work on offline. Here's an Android demo built with inception model for image identification. This app can identify simple basic objects such as notebook, mouse, water bottle, phone, books, pen, table chair, etc look at the top texts of the photo it describes the object.
There are other open-source models for natural language processing, computer vision, image captioning, generating words, text recognition or else you can train and generate your models to use with Tensorflow for your own very need.

Core ML

Core ML is a new foundational machine learning framework used across Apple products, including Siri, Camera, and QuickType. Core ML delivers blazingly fast performance with easy integration of machine learning models enabling you to build apps with intelligent new features using just a few lines of code.

Interestingly you can train your models in Tensorflow and convert them to the Core ML format so you can take advantages of both platforms. Here's a nice list of pre-trained models for Core ML and a good tutorial to learn it.

Wit.ai

Wit.ai is a natural language processing platform that makes it possible for developers to add intelligent conversation functionality to applications. Developers can use the Wit.ai API to add an intelligent voice or chat interface to home automation, connected car, smart TV, robotic, smartphone, wearable, and many other types of applications or you can build an intelligent bot and got your application connected to the bot via API. Users can enjoy a hands-free mobile experience while driving, working out, cooking.
The basic idea of integration of Wit.ai to your mobile app is via API like node, python, ruby. There are many recipes out there for you. API training is done around Stories (domain-specific use cases), where the engine learns conversation flow from examples of user input with bot response.

This is a simple android app connected with Wit.ai, it takes images as input and describes what is in it in natural language. You can build an app that can talk to you. Great!

Here's another example of Wit.ai. Users can order pizza and it can have a smart conversation with the users. It can be a nifty feature in your mobile app.

IBM Watson

IBM Watson is well-known platforms for machine learning with cognitive computing. It provides APIs such as speech to text, text to speech, trade-off analytics, personality insights, question and answer, tone analyzer, and visual recognition. You can build very smart applications with these APIs without banging your heads! everything is ready at your hands. Take a look at "10 IBM Watson-powered Apps That Are Changing Our World" and this one too "5 unusual things you can do with IBM's Watson".

Google Cloud Machine Learning Platform

Google Cloud ML Platform provides modern machine learning services, with pre-trained models and a service to generate your tailored models. Major Google applications use Cloud Machine Learning, including Photos, the Google app, Translate and many other.
Google itself uses this service to develop its smart applications. Usage of their service is massive you can do a lot of things in your applications with the services they provide, and their documentation is mind-blowing.

Amazon AI

Amazon AI is a machine learning service provided by Amazon. Under Amazon AI umbrella there is Rekognition, a service that makes it easy to add image analysis to your applications. With Rekognition, you can detect objects, scenes, and faces in images. You can also search and compare faces. Rekognition's API enables you to quickly add sophisticated deep-learning-based visual search and image classification to your applications.

Rekognition identifies thousands of objects such as vehicles, pets, or furniture, and provides a confidence score. Rekognition also detects scenes within an image, such as a sunset or beach. This makes it easy for you to add features that search, filter, and curate large image libraries.
It has other capabilities like Facial Analysis as an example it can locate faces within images and analyze face attributes, such as whether or not the face is smiling or the eyes are open, Face Comparison like measures the likelihood that faces in two images are of the same person, and Facial Recognition. There are many use cases for this service like Searchable Image Library, Face-Based User Verification, Sentiment Analysis.

Another service under Amazon AI is Polly a text-to-speech service that uses a lot of machine learning smarts under the hood.
Amazon Polly is a service that turns text into lifelike speech. Polly lets you create applications that talk, enabling you to build entirely new categories of speech-enabled products.
The third and probably most important new service is called Lex. It is essentially the technology that fuels Amazon's own Alexa service. It allows you to build conversational applications that can feature multi-step conversations.
You can use Amazon Lex to build chatbots and mobile applications that support engaging, lifelike interactions. It has many interesting use cases like Informational Bots, Application Bots like Alexa.

There are other services out there you can explore like Microsoft Cognitive Services, AlchemyAPI.

Before finishing we all know machine learning stuff is really cool but we shouldn't implement it in our app just because it is cool, if scenario really calls for it then we should go for it, like you want to make an app like Uber now you need to fraud analysis then you can go for like the Uber did so, or like Google's Allo app they extensively used natural language processing for their Google Assistant.
Trends of intelligent apps are going on, time to catch up. :) Happy Learning.

Did you find this article useful? Which technology do you want to try out among them? Leave a comment to discuss it!
If you enjoyed the writings then please use the ❤ heart below to recommend this article so that others can see it.
Let's connect through LinkedIn.

A Caching Library, written in Swift that can cache Anything.

Sadman Samee — Sun, 08 Sep 2019 16:55:11 +0000

Needed a cache Library that has

[x] Asynchronous data downloading and caching.
[x] Asynchronous image downloading, caching and showing.
[x] Expiry date/time for all the objects individually.
[x] Multiple-layer hybrid cache for both memory and disk.
[x] Fine control on cache behavior. Customizable expiration date and size limit.
[x] Force refresh if needed.
[x] Independent components. Use the Cachy or CachyLoader system separately as you need.
[x] Can save JSON, UIImage, ZIP or anything.
[x] View extensions for UIImageView.
[x] Indicator while loading images.

Couldn't find any all in one solution, either it's too much or too less, so came up with this.

https://github.com/Sadmansamee/CachyKit

it's the first version there might be a lot of issues or bad, please point them out and how I can improve them.

Working with WebRTC

Sadman Samee — Tue, 20 Aug 2019 09:48:55 +0000

Anybody who worked with WebRTC will agree that there is a huge lack of learning resources and on top of that on every few years, WebRTC standards changes and changes are big sometimes that makes previous resources outdated, creating more gap.

WebRTC is a free, open project that provides browsers and mobile applications with Real-Time Communications (RTC) capabilities via simple APIs. The WebRTC components have been optimized to best serve this purpose.

I have found most of the resources either not clear, not compatible with the current WebRTC library or very incomplete. During my searching for resources, I have found a few as good resources.

Web

Android

iOS

Flutter

These are very high-level examples, I have a plan to write a full-fledged tutorial on this topic soon.

SwiftUI Cheat Sheet

Sadman Samee — Mon, 10 Jun 2019 14:25:51 +0000

SwiftUI Cheat Sheet

This was originally published in here, in Github

SwiftUI Cheat Sheet
- Table of Contents
Tutorial
UIKIT equivalent in SwiftUI
View
- Text
- Image
- Shape
Layout
- Background
- VStack
- HStack
- ZStack
Input
- Toggle
- Button
- TextField
- Slider
- Date Picker
- Segmented Control
- Tap
- Gesture
List
Containers
- NavigationView
- Group
Alerts and Action Sheets
Navigation
Work with UIKIT
- Navigate to ViewController
- Use UIKit and SwiftUI Views Together

Tutorial

UIKIT equivalent in SwiftUI

UIKIT	SwiftUI
UILabel	Text
UIImageView	Image
UITextField	TextField
UITextView	No equivalent (use Text)
UISwitch	Toggle
UISlider	Slider
UIButton	Button
UITableView	List
UICollectionView	No equivalent (can be implemented by List)
UINavigationController	NavigationView
UIAlertController with style .alert	Alert
UIAlertController with style .actionSheet	ActionSheet
UIStackView with horizontal axis	HStack
UIStackView with vertical axis	VStack
UISegmentedControl	SegmentedControl
UIStepper	Stepper
UIDatePicker	DatePicker
NSAttributedString	No equivalent (use Text)

View

Text

To show a text in UI simply write



Text("Hello World")

To add style



Text("Hello World")
    .font(.largeTitle)
    .foregroundColor(Color.green)
    .lineSpacing(50)
    .lineLimit(nil)
    .padding()

To format text inside text view



static let dateFormatter: DateFormatter = {
    let formatter = DateFormatter()
    formatter.dateStyle = .long
    return formatter
    }()

    var now = Date()

    var body: some View {
        Text("Task due date: \(now, formatter: Self.formatter)")
    }

Image

To show image



Image("hello_world") //image name is hello_world

To use system icon



Image(systemName: "cloud.heavyrain.fill")

you can add style to system icon set



Image(systemName: "cloud.heavyrain.fill")
    .foregroundColor(.red)
    .font(.largeTitle)

Add style to Image



Image("hello_world")
    .resizable() //it will sized so that it fills all the available space
    .aspectRatio(contentMode: .fill)
    .padding(.bottom)

Shape

To create Rectangle



Rectangle()
    .fill(Color.red)
    .frame(width: 200, height: 200)

To create circle



Circle()
    .fill(Color.blue)
    .frame(width: 50, height: 50)

Layout

Background

To use image as a background



Text("Hello World")
    .font(.largeTitle)
    .background(
        Image("hello_world")
            .resizable()
            .frame(width: 100, height: 100))

Gradient background



Text("Hello World")
    .background(LinearGradient(gradient: Gradient(colors: [.white, .red, .black]), startPoint: .leading, endPoint: .trailing), cornerRadius: 0)

VStack

Shows child view vertically



VStack {
    Text("Hello")
    Text("World")
}

Styling



VStack (alignment: .leading, spacing: 20){
    Text("Hello")
    Divider()
    Text("World")
}

HStack

Shows child view horizontally



HStack {
    Text("Hello")
    Text("World")
}

ZStack

To create overlapping content use ZStack



ZStack() {
    Image("hello_world")
    Text("Hello World")
        .font(.largeTitle)
        .background(Color.black)
        .foregroundColor(.white)
}

Input

Toggle

Toggle lets users move between true and false states



    @State var isShowing = true //state

    Toggle(isOn: $isShowing) {
        Text("Hello World")
    }.padding()

Button

To create button



Button(action: {
    // do something
}) {
    Text("Click Me")
}

To create image Button



Button(action: {
    // do something
}) {
    Image("hello_world")
}

TextField

It heavily relies in state, simply create a state and pass it as it will bind to it



@State var fullName: String = "Joe" //create State

TextField($fullName) // passing it to bind
    .textFieldStyle(.roundedBorder) //adds border

To create secure TextField



@State var password: String = "" //create State

SecureField($password) // passing it to bind
    .textFieldStyle(.roundedBorder) //adds border

Slider



@State var value: Double = 0 //create State

Slider(value: $value, from: -100, through: 100, by: 1)

Date Picker



@State var selectedDate = Date()
DatePicker(
            $selectedDate,
            maximumDate: Date(),
            displayedComponents: .date
        )

Segmented Control



@State var favoriteColor = 0
var colors = ["Red", "Green", "Blue"]

SegmentedControl(selection: $favoriteColor) {
    ForEach(0..<colors.count) { index in
        Text(self.colors[index]).tag(index)
    }
}

Tap

For single tap



Text("Tap me!")
    .tapAction {
       print("Tapped!")
}

For double tap



Text("Tap me!")
    .tapAction (count: 2){
       print("Tapped!")
}

Gesture

Gesture like TapGesture, LongPressGesture, DragGesture



Text("Tap")
    .gesture(
        TapGesture()
            .onEnded { _ in

            }
        )

Text("Long Press")
    .gesture(
        DragGesture(minimumDistance: 50)
            .onEnded { _ in

            }
        )

Text("Drag Me")
   .gesture(
        LongPressGesture(minimumDuration: 2)
            .onEnded { _ in

            }
        )

List

To create static scrollable List



List {
    Text("Hello world")
    Text("Hello world")
    Text("Hello world")
}

To create dynamic List



let names = ["Thanos", "Iron man", "Ant man"]
List(names) { name in
        Text(name)
    }

To add section



 List {
    Section(header: Text("Good Hero")) {
        Text("Thanos")
    }

    Section(header: Text("Bad Heros")) {
        Text("Iron man")
    }
}

To make it grouped add .listStyle(.grouped)



 List {
    Section(header: Text("Good Hero")) {
        Text("Thanos")
    }

    Section(header: Text("Bad Heros")) {
        Text("Iron man")
    }
}.listStyle(.grouped)

Containers

NavigationView

NavigationView is more/less like UINavigationController, It handles navigation between views, shows title, places navigation bar on top.



NavigationView {
    Text("Hello")
        .navigationBarTitle(Text("World"), displayMode: .inline)
}

For large title use .large

Add bar items to NavigationView



NavigationView {
    Text("Hello")
        .navigationBarTitle(Text("World"), displayMode: .inline)
        .navigationBarItems(trailing:
                Button(action: {
                    print("Going to Setting")
                }) {
                    Text("Setting")
                })
}

Group

Group creates several views to act as one, also to avoid Stack's 10 View maximum limit.



 VStack {
        Group {
            Text("Hello")
            Text("Hello")
            Text("Hello")
        }
        Group {
            Text("Hello")
            Text("Hello")
        }
    }

Alerts and Action Sheets

To Show an Alert



Alert(title: Text("Title"), message: Text("message"), dismissButton: .default(Text("Ok!")))

To Show Action Sheet



ActionSheet(title: Text("Title"), message: Text("Message"), buttons: [.default(Text("Ok!"), onTrigger: {

    })])

Navigation

Navigate via NavigationButton



NavigationView {
    NavigationButton(destination: SecondView()) {
        Text("Show")
    }.navigationBarTitle(Text("First View"))
}

Navigate via tap on List Item



let names = ["Thanos", "Iron man", "Ant man"]
List(names) { name in
    NavigationButton(destination: HeroView(name: name)) {
        Text(name)
    }
}

Navigate via PresentationButton



PresentationButton(Text("Tap"), destination: HeroView())

Work with UIKIT

Navigate to ViewController

It's possible to work with UIKIT component from SwiftUI or call SwiftUI view as View Controller from UIKIT.

Let's say you have a View Controller named as SuperVillainViewController and want to call from SwiftUI view, to do that ViewController need to implement UIViewControllerRepresentable



struct SuperVillainViewController: UIViewControllerRepresentable {
    var controllers: [UIViewController]
    func makeUIViewcontroller(context: Context)  SuperVillainViewController {
        // you could have a custom constructor here, I'm just keeping it simple
        let vc = SuperVillainViewController()
        retrun vc
    }
}

Now you can use it like



NavigationButton(destination: SuperVillainViewController()) {
        Text("Click")
}

Use UIKit and SwiftUI Views Together

To use UIView subclasses from within SwiftUI, you wrap the other view in a SwiftUI view that conforms to the UIViewRepresentable protocol. (Reference)

as example




import SwiftUI
import MapKit

struct MapView: UIViewRepresentable {
    func makeUIView(context: Context) -> MKMapView {
        MKMapView(frame: .zero)
    }

    func updateUIView(_ view: MKMapView, context: Context) {
        let coordinate = CLLocationCoordinate2D(
            latitude: 34.011286, longitude: -116.166868)
        let span = MKCoordinateSpan(latitudeDelta: 2.0, longitudeDelta: 2.0)
        let region = MKCoordinateRegion(center: coordinate, span: span)
        view.setRegion(region, animated: true)
    }
}

struct MapView_Preview: PreviewProvider {
    static var previews: some View {
        MapView()
    }
}

Keep your iOS projects healthy

Sadman Samee — Mon, 08 Apr 2019 17:18:59 +0000

Keeping your code and projects is the best thing you can do to your projects and here’s how to do it your iOS projects written in swift.

Following style guide

There are several style guide for writing swift, and every team member should follow them well. Such as

Using tools

In reality not everyone does it properly so there should be a way to enforce it amongst team members, a way of doing is manual code review which is very time-consuming and code review should happen in logic level not how a developer should declare variable or indentation, we could simple automate these things and invest our time more on harder things.

Swiftlint will excellently lint the swift codes.
You can also add linting for storyboard and xib files with xiblint
It is also preferred to have a one way of formatting code like how many indents, way declaring variable etc and this is where Swiftformat shines really well.
To Eliminate Unused Swift Code go for periphery
Having a static analysis of your codes is a good way to have fewer bugs, Infer is a static analysis tool that produces a list of potential bugs.
You can also automate using these tools on Continuous Integration level using danger. You can use Danger to codify your teams’ norms. Leaving humans to think about harder problems. And danger supports Circle, Travis, Jenkins, Buildkite, BuddyBuild, Semaphore, TeamCity, Xcode Bots, Drone, Surf and Bitrise.

Organising Xcode projects

In older versions of Xcode you could only group and in beneath there was no folder structure so basically every file used to stay in one single folder by default and this is messy!

You can use Synx, a command line tool to create folder according to Xcode group.
And sometimes your Xcode project might be corrupted and broken and you can get back your Xcode project based on folder structure using XcodeGen, A Swift command line tool for generating your Xcode project.

DEV Community: Sadman Samee

Intelligence in Mobile Applications.

Tensorflow Lite

Core ML

Wit.ai

IBM Watson

Google Cloud Machine Learning Platform

Amazon AI

A Caching Library, written in Swift that can cache Anything.

Working with WebRTC

Web

Android

iOS

Flutter

SwiftUI Cheat Sheet

SwiftUI Cheat Sheet

Table of Contents

Tutorial

UIKIT equivalent in SwiftUI

View

Text

Image

Shape

Layout

Background

VStack

HStack

ZStack

Input

Toggle

Button

TextField

Slider

Date Picker

Segmented Control

Tap

Gesture

List

Containers

NavigationView

Group

Alerts and Action Sheets

Navigation

Work with UIKIT

Navigate to ViewController

Use UIKit and SwiftUI Views Together

Keep your iOS projects healthy

Following style guide

Using tools

Organising Xcode projects