Objects classification without code — How to use shared dataset and model

CIFAR 10 classification

Experience nocoding AI ➟ https://pro.nocodingai.com

Well-trained AI can discriminate what is the object. Today, we have 10 different objects(airplane, automobile, bird, cat, deer, dog, frog, horse, ship ans truck) to train AI. This is the classic problem of AI classification, known as ‘CIFAR10’.

We will start a tutorial with shared dataset and model.

Sign in NocodingAI platform(https://pro.nocodingai.com) and go to the menu ‘Dataset’. At the first time, you will see the page below. Press ‘Create New’ button and follow instructions.

After entering Dataset Name and Description, select below dataset card ‘cifar10’. It has 10,000 images of 10 objects, each object has 1,000 images.

Now you can check out the images of 10 classes. Each class is devided by folders. By clicking folder name, you can see which images in it.

Don’t be embarrased that image looks blurry. This raw image is very tiny(32 by 32 pixels).

(futher explanation — add image to dataset)

Let me show you other high-resolution image preview. From folder name, right click to add image. For example, I added a cat image which is 1,440 by 1,080 pixels.

Click the image to enlarge. The bigger the cuter. Yes, It’s my cat. Note that for the better training result, all the images should be sized equally (or similar sized image to resize). Just remove the added image than, click floppy disk icon to save.

Now, Move to Model menu. Srcoll down the ready-made model card, you can find ‘cifat10 classification’. Use the sample model.

There are connected layers. We will study later what is this layer, the way to change layers. Till now, please click the top right button ‘Start Training’.

Change the ‘OutputType’ to TEXT. Enter 32 for the ‘width’ and ‘height’. finally change ‘channel’ to 3. (If you want faster training, turn on the ‘Use GPU’. See our Cloud GPU subscription plan at MYPAGE)

Done! Press ‘OK’ button and monitor the results and graph.

Status changes : PENDING -> PREPROCESSING -> RUNNING (by epochs)

The training is over. You can use this trained-model to build your own application in Project menu.

Next time, we will study how to upload customized datasets.

Thank you, please subscribe this series.