Included model
Start with a bundled YOLO11 Nano ONNX model without an API key.
OML / 005 · Public release
CVD
A local workspace for computer-vision development.
CVD combines a React control surface with a Python inference service. It includes a small YOLO11 Nano ONNX model and accepts compatible ONNX or PyTorch models.
05A Model registry, input controls, inference viewport, and detection results.
01 Why it exists
Keep the complete vision loop in one place.
Model selection, device placement, input review, confidence tuning, result inspection, and training are often split across scripts and notebooks. CVD puts those tasks in one local interface.
The service binds to the local computer by default. Models and input data remain on the machine unless the operator changes the network configuration.
02 The system
Flexible inputs
Use images, video files, or a camera stream for local inference.
Device control
Select an available CPU, GPU, or Apple acceleration path.
Visual results
Review detections, confidence values, classes, and performance data.
Training workspace
Upload a YOLO data set and monitor a local training job.
Export tools
Prepare supported models for Core ML or TensorFlow Lite workflows.
03 Training view
Source available
Bring your own model.
CVD uses the PolyForm Noncommercial License 1.0.0. The repository includes setup instructions, security limits, and the demonstration ONNX model.
Get CVD ↗