Introduction to google teachable machine 2.0
Google teachable machine is an interesting AI model activity. In this post, we will talk about google teachable machine 2.0. Before you start reading, I would like to recommend you to go through AI Domains and return back to go ahead. Before starting this you need to understand machine learning. So lets we talk about machine learning. Read the complete post so you can enjoy the reading and learning.
There are two words in Machine Learning. As one of the superhit Bollywood movie 3 idiots states that machine can be anything which makes our work simple. Machines are generally operated by fuel or electricity or power. They can perform our tasks in an easy manner and make feel is comfortable about doing work. They do complex tasks very fast and reliable.
Learning means the ability to acquire a skill. Learning can be also related to grasped, knowledge or information or training too. So machine learning can be like this the complex task done by machines after training. Machines can be trained to recognise or complete the task in Machine Learning. So I hope you are clear with these words.
So google teachable machine 2.0 is a tool on the internet that makes Machine Learning models comfortable, easy to understand and use. It also generates the code for a different platform to embed in different platform. It allows creating the model and train the model for identification or recognition.
There are three basic steps of using google teachable.
- Gather: Collect the images by upload or live recording
- Train: Training the model
- Export: Export code for a specific platform
Types of Projects
Google teachable offers 3 types of projects to train the model.
- Image Project: Works on image
- Audio Project: Works on project
- Pose Project: Works on the image again but records the pose like standing pose, sitting pose etc
Let’s see how to work on Google Teachable Machine 2.0
Click here to launch google teachable. You will get the following screen.
Selecting Project type
Click on anyone get started button. A new screen will appear like this:
Uploading or recording images
Now select Image project, you will get this screen. It will also train you about how to start and train the model step by step you can follow that also.
If you have a webcam then use the webcam to store the images for training. I have done with image upload as I don’t have a webcam on my PC. So I have clicked on upload button. In my project, I am using the training model for recognizing animals like cat, dog and monkey. So first I will gather all images one by one when I have clicked on the upload button, it opened following screen.
Rename the class 1 by clicking on the pencil icon on top and give a new name cat. Now click on choose images from your computer files or drag & drop here button or you can upload images from cloud google drive. I have added 10 different images of cats through the upload button.
Now scroll down and dot the similar activity to gather dog images.
Add a class
Now click on + Add a Class button to upload monkey images.
Upload the images in the same manner.
Traning the model
Now the third step is going to start. Click on Train Model button. It will display a message like preparing training data under the Train Model button. Wait for some time. When this process is going on on your screen do not press any key or switch over any tab. When it finishes it will convert the button text into model trained. Now new part will open for preview. Kindly ensure the input must be on and file must be selected for uploaded image project.
Check the Result
Here I am uploading cat image first which is not in our uploaded images. and it will check and recognize the image. It shows the result in the percentage of matching uploaded image with a resultant image. In my result, it is showing 100% match result.
Now lets we check result for a dog image. I have selected a puppy image to check the image variations. It’s showing 79% matching with a g and rest with cat and monkey. See in below screenshot:
Now let’s have look at monkey image result.
To save the project in your computer, click on the menu at the left corner on google teachable machine 2.0. A drop will appear as shown in the following screenshot:
Click on download project as file option and save in your computer.
Always remember that the sharable link option is always just copying code, not the trained model. So those images will be uploaded again on the machine where the model implemented or used.
Open existing project:
To open any existing trained model follow these steps:
Step 1. Click on left side top menu. A drop-down menu will appear.
Step 2. Click on the open project from file button.
So I hope you have enjoyed this activity. Now you can do the projects with sound and pose in the same manner and see the result of google teachable machine 2.0 and see how accurate it is!!! Share this post with your classmates or friends and feel free to comment on your doubts and feedback. Your comments motivate us to provide information like this to you about new topics like this with the curriculum.