can someone provide a proper path.
How do I develop a model for deep fake image detection? I have almost finished the deeplearning course and yet i am not sure I can develop a functioning model without instructions. How do I approach a deep learning problem and come up with a unique solution?Do I use pre-existing model?If yes then how do I alter them for my model and how do I improve accuracy? If no then how do I design my own model.
How do I choose a model ? Is it random guessing or do we have a criteria for choosing and making a model perform better?
You have to be a vigilant searcher (is that a word?). There’s tons of documentation out there, that’s one of many great things about python. There is so much out there, if you can imagine a project, someone has probably done it already.
You can start on the open source project website. If you go to tensorflow or SKLearn or any major python package, they will have guided tutorials and walk you through all the functions with code examples.
I would also try medium.com
This is a place where data scientists blog and post details about their projects including code.
Here is an article on deepfake image detection using tensorflow.
I believe you have to be a paid member to see the data scientist blog posts.
I think it’s best to keep accurate perspective on how much complexity you are getting into. Machine Learning is very complex, takes 6 - 12 months to build core skills.
I recommend focusing on building core skills. You’ll be able to do interesting projects on your own.
1 Like