Problem with snapshots and training.

Hello,

I find myself with several problems for my model.
I have an image collection of 747 images but I can’t create a snapshot of 10 000 images.
Then with my snapshot of 747 images, I can’t make a 5h training. It stops only after 22 minutes.

It appears that your image collection is a “standard” collection type. This means that each image in the collection will appear once, as is, in the resulting dataset.

Larger datasets can be created from “Seed” image collections. Seed image collections contain images with transparent backgrounds that can be used by our Dataset Generator to create larger snapshots. You can read more about generating datasets here: https://docs.fritz.ai/dataset/generate-dataset

Finally, model training time is related to how many images your snapshot contains. If your model stops improving in accuracy on your training dataset before the budget is exhausted, we will stop training it automatically to save your hours. In your case, 747 is not very many images so training was stopped early. More images will result in longer training times.

Best,
Jameson

Thanks Jameson!
I’m going to try this out right away!