Specifying Annotation format from Python SDK

I’m following the guide here:

The app seems to have correctly guessed the format of my annotations in about 50% of the uploads in this batch, while on others the boxes just don’t match up with where they should be based on offline visualisations. I can only assume it’s a formatting translation error - boxes that are yolo format being detected as coco and not matching or something.

I can’t really go back and check every single one of 17,000 uploads so I’m going to have to delete it all and start over - how should I be setting the format correctly? is there a list of filename clues or anything somewhere? Some default format that’s easiest for me to use in python uploads?

Roboflow supports a wide variety of annotation formats for computer vision datasets. It can ingest over 25 different annotation formats and can convert (export) annotated datasets in over 20 annotation formats source. The most common formats are COCO JSON, Pascal VOC XML, and the YOLO TXT format source.

When you upload your dataset to Roboflow, it automatically detects the format of your annotations source. However, if you’re experiencing issues with incorrect format detection, it might be best to ensure your annotations are in a format that Roboflow can easily recognize.

If you’re using Python for your uploads, you might find it easiest to use a format like COCO JSON, Pascal VOC XML, or YOLO TXT, as these are common and well-supported. However, the best format to use can depend on your specific use case and the tools you’re using in your project.

If you’re still having trouble, you might find it helpful to watch this tutorial video on how to import images in one format and export them in another using Roboflow. It also covers how to create a new workspace and project, upload images and annotation files, and export the dataset in various formats.

Remember, if you’re having issues with a large number of uploads, it might be worth testing the process with a smaller subset of your data first to ensure everything is working as expected before uploading the entire batch.

Hi! I’m Roboflow’s AI powered bot. This is an automated response powered using AI with knowledge from our docs, blog, and forum. If this solves your issue, please mark it as the solution using the checkbox icon. If this doesn’t solve your issue, please follow up with what you’ve tried/why, and the community will continue to chime in as usual.

Thanks, the autodetect is correctly finding yolo txt some/most of the time, but others it’s just doing something weird and my boxes are coming out all over the image. Is there no way to say in the upload command from the python SDK “annotation_format=yolo”

I’ve run this again today and it’s a 50% success rate when uploading text files just named image(number).txt but it does better with .coco.json files.

Is there some secret to the file name? Maybe .yolo.txt or something like that?

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.