Predict

Predict with builtin or pre-trained models

Choose a framework

Choose frameworks using the dropdown in group 3 (Train/Predict). Available frameworks will depend on what is installed in the current environment.

For example in the framework below we have Cellpose, Microsam, Monai UNET, and Random Forest available on the dropdown.

Choose builtin models

To choose a builtin model go to the model drop down in the Train/Predict group and choose one of the builtin models. The below screenshot shows how this works for the Cellpose framework. The user has the choice to choose between cyto3 and tissuenet_cp3 (there are many more Cellpose models that could be added at a future date)

Load Pretrained Model

In the Train/Predict group choose Load then browse to the models directory. Models that were trained with Napari-Easy-Augment-Batch-DL will be stored in the standard format for each framework, in the projects models directory.

1. Cellpose are stored under the models directory in a file without an extension.
2. Microsam models are stored in models\checkpoints\name_of_model\best.pt
3. Pytorch and Monai UNET models are stored under the models directory in a .pt file.
4. Stardist models are stored under the models directory in a folder that contains a config.json file and a weights_best.h5 file (choose the folder in this case).

Note. Models can be loaded from any source as long as they are in the correct folder.

Predict

Hit Predict Current Image or Predict All Images then labels will be generated and added to the Predictions layer.