satsim.dataset package

Submodules

satsim.dataset.augment module

satsim.dataset.augment.augment_satnet_with_satsim(dataset, augment_satsim_params, prob=0.5, rn=0, min_snr=2.0, box_pad=10)

Augments a SatNet Dataset with SatSim images. The SatSim image is added onto the example image. New synthetic targets are appended to the bounding box annotation list.

Parameters:

  • dataset – tf.Dataset, Dataset object that returns SatNet types: image: tf.float32, bounding box annotations: tf.float32, filename: tf.string, annotation filename : tf.string

  • augment_satsim_params – dict, SatSim simulation parameters

  • prob – float, probability of augmenting the example image. default=0.5

  • rn – float, estimated read noise of sensor in photoelectrons used to estimate SNR. default=0

  • min_snr – float, minimum SNR to include synthetic targets in annotation list. default=2.0

  • box_pad – int, number of pixels to pad synthetic targets on each side. default=10

Returns:

A tf.Dataset, the mapped Dataset with SatSim augmentation