fiftyone-dataset-export

Export FiftyOne datasets by setting the dataset context, selecting a format like COCO or YOLO, specifying a destination path, and running the export. This process preserves annotations, media, and metadata for ML pipelines.

FiftyOne datasets can be converted into COCO, YOLO, VOC, CVAT, and CSV formats. These standard formats support diverse ML pipelines, enabling dataset training, sharing, and archiving workflows across various label types.

Yes, you can upload FiftyOne datasets directly to the Hugging Face Hub. The export process integrates with the Hub, generating dataset cards and optionally uploading your data while preserving annotations and metadata for end-to-end workflows.

Yes, exporting FiftyOne datasets preserves annotations, media, and metadata across supported formats. The process ensures data integrity for model training, data sharing, and archiving workflows across various label types.

The best way to share a FiftyOne dataset is exporting it to standard formats like COCO, YOLO, VOC, or CSV, or publishing directly to the Hugging Face Hub. This enables interoperability across diverse ML pipelines and data sharing workflows.