Menu
Onepanel logoGetting StartedUser GuideOperator ManualReleasesRoadmap
GitHubTwitter
Onepanel logo
  • Getting Started
  • User Guide
  • Operator Manual
  • Releases
  • Roadmap
  • GitHub
  • Twitter
  • User guide
  • Workspaces
    • Launching a Workspace
    • Pause and resume
    • Switching node pool
    • Terminating a Workspace
    • Workspace Templates
  • Workflows
    • Executing a Workflow
    • Templates and parameters
    • Working with artifacts
    • Creating a Workflow Template
    • Persisting training metrics
    • Accessing TensorBoard
    • Hyperparameter tuning
    • Troubleshooting Workflows
  • Inference APIs
    • Overview
    • Create with Web UI
    • Create with Workflow Task
    • Create with Python SDK
  • Built-in and custom Workflows
    • Model training
    • Data augmentation
  • CVAT Workspace
    • Getting started with image and video annotation
    • Training with built-in models
    • Using trained models for automatic annotation
    • Adding custom training Workflows
    • CVAT 1.6.0 Workspace
  • JupyterLab Workspace
    • JupyterLab overview
    • Using Git and browsing GitHub
    • Using TensorBoard
    • Using the built-in debugger
    • Auto completion and language features
  • Sidecars
    • Filesyncer
  • SDKs
    • Python SDK
Version: Next

Built-in data augmentation Workflows

Out of the box, Onepanel supports Albumentations for data augmentation/preprocessing in the built-in training Workflow Templates.

Number of augmentation cycles indicates the number of times to apply the transforms. Setting this field to 0, bypasses any data augmentation.

Preprocessing parameters field can contain Albumentations transforms specified in YAML format.

Edit this page
Previous
« Built-in model training Workflows
Next
Getting started with image and video annotation »
Copyright © 2021 Onepanel, Inc.