> For the complete documentation index, see [llms.txt](https://docs.roboflow.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.roboflow.com/deployment/monitoring-and-analytics/vision-events/add-images-for-training.md).

# Add Images for Training

Send images from Vision Events into a Roboflow project for training.

You can move images captured by Vision Events directly into a Roboflow project so they can be used to train or improve a model. Any predictions attached to the event are carried over as annotations when they are compatible with the target project type.

## Add Images from the Vision Events Page

1. Open **Vision Events** in your workspace and select a Use Case.
2. Click **Add Images for Training** to enter selection mode.
3. Pick the target project from the dropdown.
4. Browse events and check the images you want to include. You can:
   * Select individual images from the thumbnail strip
   * Select all images on an event with the per-event checkbox
   * Select all images on the page
5. Click **Add to Project** to submit. A progress bar shows upload status.
6. When the upload finishes, click the link in the success modal to open the target project in a new tab.

## Notes

* Only images can be added. Video clips on an event are dimmed in selection mode and cannot be selected.
* Only annotations that match the target project type are imported. Other predictions are ignored.
* Images are tagged as annotated and added to a new batch in the target project.
* The right sidebar shows which projects each image already belongs to, so you can avoid duplicates.
* When you open one of these images in the annotation editor, the "Source" section shows the Use Case it came from. Click the Use Case name to open the event in Vision Events in a new tab. You need permission to view Vision Events to see this section.
* Each image gets the event's data as [image metadata](https://docs.roboflow.com/datasets/create-and-upload/adding-data/image-metadata), so you can see it in the annotation editor and search on it. Event fields start with `vision_` (ex: `vision_event_id`, `vision_use_case`, `vision_device_id`). The event's custom metadata starts with `vision_meta_`, the image's own metadata with `vision_img_`, and operator feedback with `vision_feedback` (feedback metadata uses `vision_feedback_meta_`). Fields that already exist on an image are never overwritten.