Lobe - A Free & No-Code Machine Learning Application

Introduction

Hey, you’ll learn about an interesting free machine learning application in this article. This works without a single line of code, python, or R. No language is required and there's no need to know any ML frameworks tensor flow, Keras, etc. Even cloud providers like Azure, AWS, Google Cloud, and the Internet are not necessary

Now there's a big question in your mind... Oh Really?

Yes, it exists... And one more surprise.

What?

Do you have an idea of which company produced it?...

Even I was shocked to hear that.

This is the first time Microsoft has launched such an awesome free application. Called LOBE, this app makes machine learning easy.

What is Lobe? What can we do with it?

Lobe is a Free Desktop Application developed by Microsoft. Lobe makes everything you need for your machine learning ideas into machine learning models and you can use them in your production Apps. Whatever you want to train your machine learning model, Lobe will do it automatically without any code by ease GUI. Once you have trained your model, you can use it in your application and make your app ready.

Currently, Lobe is in the Beta version and you can do Image classification. In a future version, you can expect other architecture ML Models like object detection, Data classification, etc.

Why Lobe?

• Easy GUI, anyone can train the ML model.
• You can use your trained model on any platform.
• Train in a local machine, without any programming languages and frameworks in your PC, and no need to upload data to any cloud platforms.
• Lobe cuts down the process of machine learning into three easy steps.
  • Collect and label images.
  • Train and understand your results.
  • Then play with your model and improve it.

How it works?

Label your images

Collect snaps using your webcam, or drag and drop images from your computer into the lobe. After that, label your images to create a machine

Train your model

Once you complete Labeling to uploaded images, the lobe will start to train automatically on your computer without any configuration. You can understand the strengths and weaknesses of your model with live visual results.

Play with it

You can play and predict by uploading different images and by clicking the correct or incorrect label for the newly uploaded image, this leads to improving your model result more accurately.

Ways to use it

• A model is a collection of files that other programs or applications can load to run trained predictions. You can store these files on both the structure of your model and you can weights that are a result of training.
• You can use the model locally in your application and in most major cloud platforms AWS, Azure, and Google Cloud to create an API.
• Lobe also hosts a model as a local API to help kick-start your app development in any platform Python, C#, Java, etc.

Installation

Let’s start with how to download and install this app.

Download - Go to - https://lobe.ai/ - and click download - you should join with lobe, give your name and mail details --> then the download will start.

Run the installer you can see the below screen - Click Next

You can see the installation process:

Click on Finish

You can see the below lobe starting screen.

Read the software license and Privacy policy and click - Yes, I Agree

You can see the Menu options

Click Get Started

You can see Lobe UI, which looks elegant and simple

Import Images

Import Images to train the model and categorize each image by giving the label name (Also catch live PICS or structured folder of images)

Am going to import eight Robert Downey jr images to train my model under the Robert Downey jr Label dataset.

Label Images

Once you upload the images, you can see that all the images are unlabelled categories after the upload you must name each image with which label you want.

Give the label name to all images “Robert Downey Jr.”

Note. We need to upload a different dataset. Then Lobe will start to train the model

I am going to import six Will Smith images to train my model under the Will Smith labeled dataset.

I am just naming the project “First Project”

Train images

Once you label both datasets, the lobe will automatically start to predict. You can see this in the below image:

After clicking the train, you can see the accurate percentage difference in both datasets.

You can also see that it will predict one Will Smith image as Robert Downey Jr.

If you change that image label to Will Smith again, it will train that image to Will Smith

Play with model

I am going to play by importing other Will Smith pictures to make the prediction more accurate.

Uploading a new image:

It was showing a Robert Downey Jr. label on the image, click the incorrect and correct button

I am going to click the incorrect button.

Once you click incorrect, this image will be saved under the Will Smith dataset. Once finished, Lobe will train the model entirely and prediction may go wrong or accurate.

Based on how many images you are going to upload playing with your model will predict accurately.

I have uploaded more images and trained this model as accurate.

Click Menu Option --> and Click Export

You can export to Tensor Flow, Tensor Flow Lite, Local API

Export as TensorFlow and you can use cross-platform apps

Export as TensorFlow Lite and you can use mobile apps

Also, you can use it as a local API for any programming language.

If you reopen the lobe you can see the below image:

I hope this article will help you to learn something new about this Lobe application and its awesome features. Try using this application, maybe it will become a great ML application in the future.