Onepanel

Full stack vision AI on Kubernetes

Onepanel is a unified vision AI platform with fully integrated modules for model building, semi-automated labeling, training, pipelines and model serving.

[object Object]

Image and video annotation with automatic pre-annotation

Annotate images and video using CVAT, with object tracking and semi-automatic annotation of bounding boxes and polygon masks.

[object Object]

Deploy complete aerial imaging pipelines with photogrammetry tools

Leverage tools like WebODM with GPUs to accelerate image processing and to perform pre-annotation for object detection.

[object Object]

Build your models with GPU optimized frameworks and IDEs

Workspaces that upgrade or downgrade to and from GPUs machines, preloaded with PyTorch, TensorFlow, JupyterLab, VSCode and more.

[object Object]

Create pipelines as code for model training and data processing

Build reproducible training and data processing pipelines that can perform parallel or distributed tasks on multiple machines.

[object Object]

Track experiments and visualize model metrics

Track and visualize model metrics and expriments with TensorBoard and ModelDB, or bring your own experiment tracking tools.

[object Object]

Web-based VNC environments that you can access with your browser

Access and share tools like AirSim, Carla, Gazebo or OpenAI Gym through your browser with VNC enabled workspaces.

[object Object]

Bring your own tools as reproducible templates

Easily bring your own specialized cloud-native tools for development, annotation, pipelines and more.

[object Object]

Integrate additional environments and pipelines using APIs and SDKs

Extend Onepanel with our powerful REST APIs and SDKs to further automate your pipelines and environments.

[object Object]

Infrastructure and workflows as code

Workflows, environments and infrastructure are all defined as code and version controlled, making them reproducible and portable.

[object Object]

Powered by Kubernetes so you can deploy anywhere

Powered by Kubernetes so you can deploy anywhere Kubernetes can run.