Image Manual (How to use my .jsonl after I import them)

I had final project and the topic is plastic trash clasification.
I had already annotated plastic trash images and import the dataset into .jsonl format.

What should I do for the next step? How I import the dataset into my .ipynb ? And how to split them into training dataset and testing dataset ?

Please help me out with how to do this. Thankyou for your help.

Hi! Just to make sure I understand the question correctly: What's your goal, and what do you want to do next? Do you want to train your model? Do you want to collect more annotations with Prodigy?

Hi Ines, thankyou for your reply. My goal is build model for plastic trash clasification using CNN. Yes, I want to train my model using my annotations. I already annotated my images and import them into a dataset. So what should I do next? Thankyou

If you're done with annotation and you don't need to label anything anymore, you typically want to use the data and run a training experiment. That's all stuff that would happen outside of Prodigy or at least, isn't built in because it depends on your specific model implementation and how you want to update it.

So you can export your annotations and then update a model. Image annotations collected with Prodigy will contain a list of "spans" where each span has a "label" and "points" (the [x, y] pixel coordinates of the bounding box or shape you drew). That typically has all the information you need to update an image model.

Yes Ines, I want to run a training experiment. I already export my annotations into .jsonl format. It means I can use my annotations to build my training dataset and testing dataset ? Is that correct ? Thankyou.

Yes, exactly. Each line in your JSONL file is one example, so you could shuffle them all and then split them into training and evaluation examples, for instance.

Okay Inez, thankyou for your help. I will explore more about that. Have a great time :wink: