Is object labeling in images with Roboflow a paid service?

Are there any limits to the labeling processes when I create a dataset?

Our pricing page is here: Roboflow Pricing and Plans

My English is not good. There are things I don’t understand about the price. How many images can I upload and add tags per month with Starter?

Hey there @Recep_Ortac

With Starter plan under the current pricing, you can upload 10,000 source images. (Source images mean images that you upload directly, augmented images that are generated from version creation do not count toward this limit) By “tags” I assume you’re talking about annotations: You can label as many annotations as you want with the Starter plan within the image limits.

For others looking for the most up-to-date pricing information, please see the Roboflow Pricing and Plans page

1 Like

If I want to upload over 10,000 images? And if I want automatic tagging.

The current pricing for image overages (going over the plan limit) is currently $0.019 per image. You can do this and see the latest pricing by clicking the “Manage” button on your workspace’s Usage page.

For our Auto Label pricing, please see the pricing section near the bottom of the Roboflow Auto Label: Fast, Automatic Vision Labeling page.

Is relabeling paid for data boosting in the Starter plan? Does it deduct or require credits? Also, does the visual increase affect 10000 credits?

Could you explain what you mean by “relabeling paid for data boosting”? It costs one auto-label credit to automatically label one image.

Could you also explain what you mean by the “visual increase”?

I have tested the problem and there is no problem, but data boosting takes too long. Is this normal? I have doubled about 5000 photos to 5 times and I have been waiting for 5 hours on average.

Hey @Recep_Ortac

There are times when there’s a lot of demand and generating a version takes longer than usual. Please let us know if this happens consistently. As for generating 5x images, I usually would not recommend it since it either leads to either: overfitting or a low performing model (by training on a dataset that’s augmented so much that it isn’t representative of images the model will actually experience in real life)

Also, I’m assuming you’re referring to version generation when you say “data boosting”.

1 Like

How about 3x? Is it a problem if I turn off the screen while a data set is being created?

Although its difficult to say since it does depend on the actual images, the augmentations you use and the use case, 3x is usually a good spot.

Not at all, dataset generation happens on our backend so leaving/turning off your screen will not affect generation or training (through Roboflow Train) at all.

1 Like

Thank for help me! :heart:

1 Like

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.