# predefined_models

Python API: `luxonis_train.config.predefined_models`

Model graphs built from a name and a few parameters.

Name a model in the `model.predefined_model` section of a config. The config then appends the nodes of the model to `model.nodes`,
after the nodes that the config lists. These nodes carry the losses, the metrics, and the visualizers of the model.
`include_losses`, `include_metrics`, and `include_visualizers` control which of them the nodes keep.

Most models offer a `light` and a `heavy` variant, and some offer a `medium` one between them. `light` is the fastest and `heavy`
is the most accurate. Each model docstring lists its components and its variants, and most also list the measured throughput.

A predefined model is versioned. `version` selects the version, and `"latest"` is the default. A change that alters the graph of
an existing model goes into a new version directory. A config that pins `version` therefore still builds the same graph.

## Child Pages

 * [anomaly_detection](https://docs.luxonis.com/software-v3/ai-inference/model-source/training/luxonis-train/luxonis-train-api-reference/config/predefined_models/anomaly_detection.md)
 * [classification](https://docs.luxonis.com/software-v3/ai-inference/model-source/training/luxonis-train/luxonis-train-api-reference/config/predefined_models/classification.md)
 * [detection](https://docs.luxonis.com/software-v3/ai-inference/model-source/training/luxonis-train/luxonis-train-api-reference/config/predefined_models/detection.md)
 * [embeddings](https://docs.luxonis.com/software-v3/ai-inference/model-source/training/luxonis-train/luxonis-train-api-reference/config/predefined_models/embeddings.md)
 * [fomo](https://docs.luxonis.com/software-v3/ai-inference/model-source/training/luxonis-train/luxonis-train-api-reference/config/predefined_models/fomo.md)
 * [instance_segmentation](https://docs.luxonis.com/software-v3/ai-inference/model-source/training/luxonis-train/luxonis-train-api-reference/config/predefined_models/instance_segmentation.md)
 * [keypoint_detection](https://docs.luxonis.com/software-v3/ai-inference/model-source/training/luxonis-train/luxonis-train-api-reference/config/predefined_models/keypoint_detection.md)
 * [ocr_recognition](https://docs.luxonis.com/software-v3/ai-inference/model-source/training/luxonis-train/luxonis-train-api-reference/config/predefined_models/ocr_recognition.md)
 * [segmentation](https://docs.luxonis.com/software-v3/ai-inference/model-source/training/luxonis-train/luxonis-train-api-reference/config/predefined_models/segmentation.md)
 * [base_predefined_model](https://docs.luxonis.com/software-v3/ai-inference/model-source/training/luxonis-train/luxonis-train-api-reference/config/predefined_models/base_predefined_model.md)