Building Your Keypoint Detection Model

How Do I Build My First Keypoint Detection Model with Datature?

Watch a 5-minute introductory video to see how to train a keypoint detection model with Datature.

Video Recap

  1. Start by creating a new Project with task type Keypoint Detection.
  2. Upload your images by navigating to the Dataset tab and dragging an image or a folder of images into the upload box.
  3. Import annotation files or annotate by hand using our Annotator tool.
  4. Build training workflows using our recommendation system or customize it to suit your needs.
  5. Train with multiple GPUs and monitor training progress using a suite of tools including metrics, visualizations, confusion matrix, and raw logs.
  6. Export your model into a variety of formats including PyTorch and ONNX.
  7. Download your model for local inference, or create a deployment to host your model on our cloud servers where you can run inference via API calls.