After reviewing the lesson1 jupyter notebook, I’m able to understand the basic process of setting up a CNN for differentiating between cat and dog images. From what I understand, the process is:
- Set up the database of images in the path
- Determine the best learning rate to use for the data
- Training the neural net with the data and the learning rate
- Running the neural net and trying it out with example images
My problem is I don’t know how to put everything together for other problems. There’s some kind of a bonus challenge hosted on Kaggle about identifying the differences between different dog breeds, but I can’t seem to wrap my head around how to take what we’ve learned in lesson 1 and apply it to that. Will we be going over more examples of CNNs and hopefully how to write the code from scratch? Thanks!
@ecdrid it might not be a good idea to bring in approaches from completely different libraries - especially here in the beginner forum, that can be pretty confusing!
@vnator at the end of the last lesson I showed how to enter that dog breed competition from scratch. Try watching that video again and see if you can follow along. Let us know if you get stuck at any point, and we’ll do our best to help.
As explained by Jeremy you can follow the video lecture and start from there. But a few tips from my end.
- lesson2-image_models.ipynb from the course is a good starting point. You will have to download the dogs breed dataset and point the model to that.
- There is a separate forum thread on this in Part 1 v2 which you must follow. It will also help you to wrap your head and get the whole thing together. You can access the thread here - Dog Breed Identification challenge
Hope this helps.