Bring a public dataset from Dataset Ninja into OvalBee by its link — images and annotations included — ready for any other template to start from.
Overview
- Task: Data import
- Input: A link to a dataset page on Dataset Ninja
- Pipeline: download the dataset → collect it in an OvalBee dataset
- Output: An OvalBee dataset with the images, their annotations, splits and image-level labels
When you need it
When you want to try a pipeline on real labeled data before your own is ready, or to add a public benchmark next to it. A single link replaces downloading an archive and uploading it by hand.
What you need
The address of a dataset page on Dataset Ninja, for example https://datasetninja.com/pcb-component-detection. A direct link to a Supervisely or image archive works too. Pick a destination dataset or type a new name to create it.
How it works
| Step | What happens |
|---|---|
| Import | The dataset behind the link is downloaded with its annotations and added to the destination dataset |
| Dataset | The same dataset, ready to be picked by any other template |
Choose between the full dataset and its published sample. If the dataset has no sample, the import stops instead of downloading the full one.
What an import gives you
A labeled OvalBee dataset that the training, evaluation and data-quality templates can start from right away.
Related templates
- Dataset quality control — see what the imported dataset contains before you use it
- Train EdgeCrafter Detector on Images — fine-tune a detector on the imported labels
- Import images by drag & drop — upload your own images from your computer instead