# AI Models

## What AI models can run on OAK cameras?

OAK cameras run object detection, segmentation, pose estimation, classification, and other neural networks, including YOLO-family
models. Pretrained models are available in the Luxonis Model Zoo on Hub AI. Custom models trained in PyTorch, TensorFlow, or ONNX
can be converted for RVC2 or RVC4 through Hub AI, or offline with ModelConverter.

Learn more: [Luxonis Model Zoo](https://docs.luxonis.com/software-v3/ai-inference/model-source/zoo.md)

## How do I convert a custom model for an OAK camera?

Convert custom models for OAK cameras with Hub AI, the recommended online converter, or offline with Luxonis ModelConverter.
Export the trained model to ONNX first, then convert it for the target platform: RVC2 (OAK) or RVC4 (OAK4). The converted model
loads into a DepthAI pipeline through the `NeuralNetwork` or `DetectionNetwork` node.

Learn more: [Convert with Hub AI](https://docs.luxonis.com/software-v3/ai-inference/conversion/rvc-conversion/online/hubai.md) ·
[ModelConverter (offline)](https://docs.luxonis.com/software-v3/ai-inference/conversion/rvc-conversion/offline/modelconverter.md)

## Can I run YOLO on an OAK camera?

Yes. OAK cameras run YOLO object detection models on the device. Pretrained YOLO models are in the Luxonis Model Zoo, and Hub AI
converts custom-trained YOLOv5, YOLOv6, YOLOv7, YOLOv8, YOLOv9, YOLOv10, YOLO11, YOLO12, YOLO26, YOLOE, and Gold-YOLO models.
DepthAI detection nodes decode YOLO output on the camera, and on stereo models they can return 3D coordinates for each detection.

Learn more: [Model preparation](https://docs.luxonis.com/cloud/hubai/model-registry/preparation.md) · [SpatialDetectionNetwork
node](https://docs.luxonis.com/software-v3/depthai/depthai-components/nodes/spatial_detection_network.md)