NOTE: Since IBM will be removing the Visual Recognition service in Dec 2021, they have somewhat hidden the ability to add this service. Here is a direct link you can use to add the Visual Recognition service:https://cloud.ibm.com/catalog/services/visual-recognition(Links to an external site.)
After adding it via the direct link, you can then skip steps 22-26 in the lab directions.
IBM Watson Visual Recognition- Assignment 4 (50 points) IBM Watson Visual Recognition is a service that uses deep learning algorithms to identify objects, faces and other contents in an image. You can sign up if you do not have an account by clicking the following link. If you have an account you can directly login using IBM Cloud & go to the Watson Studio from Services (if you already created one) otherwise use the following link to register 1. https://cocl.us/CV0101EN_IBM_Cloud_Registration 2. On the signup page, select where you want your data to be stored in, in my case it is Dallas. Enter your email address, then check the checkbox that is surrounded by the red rectangular border indicated in the picture below. Then click the Next button. 3. After clicking Next you should end up on a page similar to the picture below. If you already have an IBM Cloud account you can just log in. But if you do not have an IBM Cloud account, you can fill in your profile information, such as your Email, First Name, Last Name, Country or Region, and set your Password. Then click on the Create Account button to create your IBM Cloud account. 4. After this you will see the below image. In a new tab, open and login to your email account https://cocl.us/CV0101EN_IBM_Cloud_Registration 5. Open the IBM Cloud confirmation email and click on Confirm Account. 6. Congratulations, you have successfully activated your IBM Cloud account. Upon clicking Confirm account you will see the following screen (i.e. you are automatically logged in). If you are automatically logged in the manner below, click on on Get Started and then click Create a Project. (Skip to Step 10) 7. OR the following screen will show where you have to manually type in your username and password to login 8. After logging in, search for Watson Studio and click on it. You will then see 9. Upon clicking it, you will see the following screen. Click on Create After clicking on Create, you will see the following screen. Click on Get Started 10. After clicking Get Started you will see the following screen 11. Click on Create a Project 12. You will then see the following screen. Click on Create a empty project 13. Give your project a name and select the following checkbox 14. Scroll down and click on Add 15. A new tab will open which will prompt you to add a Cloud Object Storage instance to your project. To make sure you don't already have one in your account do the following. Click on existing and then click on Resource Group. Select ALL checkboxes from the dropdown. 16. Similarly, click on Location and select ALL checkboxes from the dropdown 17. If an existing Cloud Object Storage instance shows up, select it and skip to Step 20. If not, proceed to the next step. 18. Click on New on the top left corner beside Existing. Upon clicking it you will see the following screen. Scroll to the bottom and select the Lite Plan. Then click on Create 19. Upon clicking Create you will see the following screen. Click Confirm 20. After this the new tab will close and you will return to your Project tab. Here click on Refresh 21. After clicking Refresh, it will show you the Cloud Object Storage Instance name that you either just created with the Lite plan or will show the name of an existing one and if you selected that one for this project. Click Create 22. After clicking Create you will see the following screen. Click on the X to close the popup 23. Click on Add to Project 24. Select the visual Recognition model 25. After selecting it, you will see the following screen prompting you to add a Visual Recognition service to your project. Click on here 26. After clicking here, you will see the following screen. Repeat steps 15 - 19 but this time we are doing it for the visual recognition service. That is, click on Existing under the Visual Recognition service, select ALL checkboxes under the Resources and Location dropdown menus and check if an existing Visual Recognition service shows up. If it does select it, else switch to New under the Visual Recognition Service, select the Lite Plan and click Create. After clicking Create, you will see a popup. Click Confirm on it. After performing these steps, you will see the following screen. You click on Custom Models → Create Model & you would see the screen below. You can also click on your Associated Service name under Default Custom Model to see the list of available classifiers For the custom model that we are creating in the visual recognition service, we are classifying 3 breeds of dogs, Husky, Beagle and GoldenRetriever. The dataset is given in the links below: NOTE: These links have more than 10 images so you may want to just keep 10 in each to speed up the process else it may take quite a while to upload those images & train a model on all these (you may not have that much time). You can also use your own images, if you have a collection of your pet(s) images. • https://cocl.us/CV0101EN_Coursera_Beagle • https://cocl.us/CV0101EN_Coursera_GoldenRetriever • https://cocl.us/CV0101EN_Coursera_Husky So let's start by creating our first class for training. Let's start with Husky, so we name our first class Husky. Let’s upload our training dataset for Husky, click on Browse, and select the Husky.zip file which contains the training images for Husky. https://cocl.us/CV0101EN_Coursera_Beagle https://cocl.us/CV0101EN_Coursera_GoldenRetriever https://cocl.us/CV0101EN_Coursera_Husky To add the uploaded dataset to our model, click on the checkbox next to Husky.zip and then click on Add to Model. Now do the same with the dataset for Beagle and GoldenRetriever. Then click Train Model. Now that your model is trained click on Trained. To test our model click on the Test option, then browse to upload your test dataset. Once you uploaded your images, Watson Studio Visual Recognition will give each image an associated tag of a class that the image falls under. Beside each tag there is a confidence score of how sure the model thinks that the image matches the tag. Deliverable: Upload the final result of your test images dataset as shown above as a doc/pdf on Canvas. Make sure you use your own images for testing purposes, so your results will vary as you use a different dataset.