Roboflow + SageMaker Ground Truth
Upload the output manifest from a SageMaker Ground Truth labeling job to Roboflow, then train, convert or relabel the dataset without a conversion script.
How to connect SageMaker Ground Truth to Roboflow
4 stepsDownload the output manifest
Ground Truth writes the labeling job's output manifest to the S3 location set when the job was created. Download it together with the images it references.
Create a Roboflow project
Create a project with the same task type as the labeling job, for example object detection for a bounding box job.
Upload the images and manifest
Drag the images and the manifest into the upload area. Roboflow parses the manifest and attaches each label to its image.
Generate a version
Click Generate New Version and apply any preprocessing or augmentation. From here you can train a model or export the version in another format, such as COCO JSON.
What teams build with SageMaker Ground Truth
Move labeled data off Ground Truth
Bring existing labeling job output into Roboflow before your team stops using Ground Truth.
Convert a manifest to COCO or YOLO
Upload a manifest once and export the same labels in the format your training code expects.
Write a manifest for Rekognition
Export a Roboflow dataset as a Ground Truth manifest, the format Rekognition Custom Labels imports.
Questions
No. AWS says SageMaker Ground Truth is no longer open to new customers; existing customers can keep using it, but no new features are planned.
The output manifest from the labeling job, a JSON Lines file with one image per line, plus the images it points to.
Yes. Sagemaker GroundTruth Manifest is one of Roboflow's export formats, so you can produce a manifest from any dataset version.
Why are companies choosing Roboflow?
The end-to-end platform for vision AI: create datasets, train models and deploy to production.
Flexible deployment modes
Move from a scalable cloud API to edge hardware like NVIDIA Jetson without managing complex dependencies.
Integrated workflow builder
Combine multiple AI models, custom business logic and third-party integrations into one production-ready pipeline in a low-code editor.
Model monitoring and alerts
Dashboards track confidence, inference time and request volume for your deployed models, and email alerts tell your team when confidence suddenly drops.
