Reviewers are not looking for the most complicated project. They want to see that you understood a real problem, chose a sensible approach, and can explain your decisions.
Start with a problem you have seen yourself — at college, at home, in your town. Then check that there is a dataset or hardware you can actually get.
Keep the scope tight. A project that does one thing well beats one that does five things badly.
Finally, practise explaining it. Be ready to answer why you chose this algorithm, what the accuracy means, and what you would improve next.