EffNetImageEmbedder

class lightautoml.image.image.EffNetImageEmbedder(model_name='efficientnet-b0', weights_path=None, is_advprop=True, device=torch.device)[source]

Bases: torch.nn.Module

Class to compute EfficientNet embeddings.

__init__(model_name='efficientnet-b0', weights_path=None, is_advprop=True, device=torch.device)[source]

Pytorch module for image embeddings based on efficient-net model.

Parameters

model_name (str) – Name of effnet model.
weights_path (Optional[str]) – Path to saved weights.
is_advprop (bool) – Use adversarial training.
devices – Device to use.

get_shape()

Calculate output embedding shape.

Return type: int
Returns: Shape of embedding.