Available pre-trained models

The following pre-trained models are available (for a more detailed description, please refer to the Methods section of the paper as well as Supplementary Table 2):

  • general: trained on a diverse dataset of spots of different modalities acquired in different microscopes with different settings. This model is the default one used in the CLI (pixel sizes: 0.04µm, 0.1µm, 0.11µm, 0.15µm, 0.32µm, 0.34µm).

  • hybiss: trained on HybISS data acquired in 3 different microscopes (pixel sizes: 0.15µm, 0.32µm, 0.34µm).

  • synth_complex: trained on synthetic data, which includes simulations of aberrated spots and fluorescence background (pixel size: 0.1µm).

  • synth_3d: trained on synthetic 3D data, which includes simulations of aberrated spots and Z-related artifacts (voxel size: 0.2µm).

  • smfish_3d: fine-tuned from the synth_3d model on smFISH 3D data of Platynereis dumerilii (voxel size: 0.13µm (YX), 0.48µm (Z)).

You can use these models to predict spots in images or to fine-tune them on a few annotations of your own data. The models can be loaded via the API as follows:

from spotiflow.model import Spotiflow

pretrained_model_name = "general"
model = Spotiflow.from_pretrained(pretrained_model_name)

You can also load them from the napari plugin or from the CLI by specifying the name of the model. See Predicting spots using the napari plugin and Inference via CLI for more information respectively.