Picking your Robotic Inference Idea

Project guidelines:

In addition to training a network on the supplied data, you will also need to choose and train a network using your own collected data. At a minimum, it needs to be a classification network with at least 3 classes. However, if you would like to be more adventurous you can use more than 3 classes and even subclasses!

If you are looking for an extra challenge, you can create a detection network. It will require you to annotate your data in addition to collecting it. More information can be found in the next section on this process.

Its okay to use a sample idea below if you’re having a hard time deciding what to do!

Resources and ideas:

• Pill identifier with classes: (pill a, pill b, pill c, no pill)
• Defective item vs normal item with classes: (no item, defective item, normal item)
• Person vs no person with classes: (correct person, wrong person, no person)
• Location of robot part on a workbench.
• Insert your idea here!

Fun examples to check out!

~Blackjack

More to come soon!