adam.model.infer_lake_breeze_batch#

adam.model.infer_lake_breeze_batch(radar_list, model_name='lakebreeze_best_model_fcn_resnet50', area_threshold=20)[source]#

This module will infer the location of the lake breeze from a batch of radar images.

Parameters:

  • radar_scan (list of RadarImage) – The list of RadarImages containing the radar image.

  • model_name (str) – The model to use. Currently, ADAM has 2 models: lakebreeze_model_fcn_resnet50_no_augmentation: A fine-tuned ResNet50 with no data augmentation. Typically more liberal in detecting lake breezes. lakebreeze_best_model_fcn_resnet50: A fine-tuned ResNet50 trained with data augmentation. Typically more conservative in detecting lake breezes.

  • area_threshold (int) – The minimum continuous area in pixels for lake breeze segments. This helps remove false positive speckles that are identified by the model.

Returns:

radar_scan – The RadarImages with the lake breeze mask.

Return type:

list of RadarImage()