core

README

Production scale, vision AI platform, with fully integrated components for model building, automated labeling, data processing and model training pipelines.

Why Onepanel?

• End-to-end automation for production scale vision AI pipelines
• Best of breed, open source deep learning tools seamlessly integrated in one unified platform
• Infrastructure automation so you can easily scale your data processing and training pipelines to multiple nodes
• Customizable, reproducible and version controlled tooling and pipeline templates
• Scalability, flexibility and resiliency of Kubernetes without the deployment and configuration complexities

Features

• Annotate images and video with automatic annotation of bounding boxes and polygon masks, integrated with training pipelines to iteratively improve models for pre-annotation and inference.
• JupyterLab configured with extensions for debugging, Git/GitHub, notebook diffing and TensorBoard and support for Conda, OpenCV, Tensorflow and PyTorch with GPU and much more.
• Build fully reproducible, distributed and parallel data processing and training pipelines with real-time logs and output snapshots.
• Bring your own IDEs, annotation tools and pipelines with a version controlled YAML and Docker based template engine.
• Track and visualize model metrics and experiments with TensorBoard or bring your own experiment tracking tools.
• Extend Onepanel with powerful REST APIs and SDKs to further automate your workflows.

Online demo

We have created an online demo environment so that you can quickly try Onepanel.

Note that this is a shared demo environment with the following restrictions:

• Data is reset every few hours
• One type of node pool (machine type) with a limit of 5 concurrent nodes
• Certain actions may be restricted

Quick start

See quick start guide to get started with the platform of your choice.

Quick start videos

Getting started with Microsoft Azure
Getting started with Amazon EKS
Getting started with Google GKE

Community

See documentation to get started or for more detailed operational and user guides.

To submit a feature request, report a bug or documentation issue, please open a GitHub pull request or issue.

For help, questions, release announcements and contribution discussions, join us on GitHub discussions.

Contributing

Onepanel is modular and consists of the following repositories:

Backend (this repository) - Code base for backend (Go)
Frontend - Code base for frontend (Angular + TypeScript)
CLI - Code base for installation and management CLI (Go)
Manifests - Kustomize manifests used by installation and management CLI (YAML)
Python SDK - Python SDK code and documentation (Python)
Templates - Various Workspace, Workflow, Task and Sidecar Templates
Documentation - The repository for documentation site
API Documentation - API documentation if you choose to use the API directly

See CONTRIBUTING.md in each repository for development guidelines. Also, see contribution guide for additional guidelines.

Acknowledgments

Onepanel seamlessly integrates the following excellent open source projects. We are grateful for the support these communities provide and do our best to contribute back as much as possible.

Argo
CVAT
JupyterLab

License

Onepanel is licensed under Apache 2.0.

Need a managed solution?

Visit our website for more information about our managed offerings.