Computer-vision project for summer intern

I am a prefinal year undergrad with decent computer vision knowledge. This summer I am planning to apply as a summer intern to various professors. Any suggestions on what I shall present as my research idea?

3 Likes

As you said thay you have descent knowledge on CV, IMO you may want to focus on GAN based models. Although, GAN itself is very vast field but i can suggest you one topic that i myself thought in my final year.

You may have heard of StyleGAN which produces very high res images of persons that doesn’t exist.

You can focus on creating faces between interpolation space of two images. There can be infinite numbers or faces generated by using linear interpolation of two face vectors. You could start with two and then expand it to work with even multiple face vectors in which the interaction of vectors may become little bit complex.

2 Likes

Okay, I’ll keep that in mind. I was thinking of training Siamese like models to identify faces even with their masks on. Biometrics with face masks. How is that?

2 Likes

That is good idea but I am not sure whether any public dataset is available for it.

1 Like

There Is GeorgiaTech:

3 Likes

I suggest to take a look to:

3 Likes

Wow, thanks a lot sir.

1 Like

If you want to explore some more advanced topics in the same subdomain you could take a look at:

Hopefully during your summer intern you could contribute a TF version in our model garden :wink:

And a dataset wrapper in:

2 Likes

Many thanks sir. I sure will take a look :slight_smile:

1 Like

Here’s a deck I put tgoether last year that might be helpful for you to come up with project ideas and work on them in systematic ways:

3 Likes

Yes! This is a great opportunity for contributing to the Model Garden repository which is under active development.

Besides model implementations (which may seem a bit daunting during first times) you can also consider contributing Colab Notebooks as tutorials which are extremely important to make Model Garden as accessible as possible to the community.

2 Likes

Thank you. This is really helpful.
(haha, coincidentally, I stumbled upon PyImageSearch just yesterday)

1 Like

Thanks for sharing the deck!!

1 Like