bacpipe.model_pipelines.feature_extractors package

Submodules

bacpipe.model_pipelines.feature_extractors.audiomae module

class bacpipe.model_pipelines.feature_extractors.audiomae.Model(**kwargs)[source]

Bases: ModelBaseClass

preprocess(audio)[source]
class bacpipe.model_pipelines.feature_extractors.audiomae.PatchEmbed_new(img_size=224, patch_size=16, in_chans=3, embed_dim=768, stride=10)[source]

Bases: Module

Flexible Image to Patch Embedding

forward(x)[source]

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

get_output_shape(img_size)[source]

bacpipe.model_pipelines.feature_extractors.audioprotopnet module

class bacpipe.model_pipelines.feature_extractors.audioprotopnet.Model(**kwargs)[source]

Bases: ModelBaseClass

classifier_predictions(embeddings)[source]
preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.aves_especies module

class bacpipe.model_pipelines.feature_extractors.aves_especies.Model(birdaves=False, nonbioaves=False, **kwargs)[source]

Bases: ModelBaseClass, Module

preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.avesecho_passt module

class bacpipe.model_pipelines.feature_extractors.avesecho_passt.AugmentMelSTFT(n_mels=128, sr=32000, win_length=800, hopsize=320, n_fft=1024, freqm=48, timem=192, htk=False, fmin=0.0, fmax=None, norm=1, fmin_aug_range=1, fmax_aug_range=1000)[source]

Bases: Module

extra_repr()[source]

Return the extra representation of the module.

To print customized extra information, you should re-implement this method in your own modules. Both single-line and multi-line strings are acceptable.

forward(x)[source]

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

class bacpipe.model_pipelines.feature_extractors.avesecho_passt.Model(**kwargs)[source]

Bases: ModelBaseClass

classifier_predictions(embeddings)[source]
preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.bat module

class bacpipe.model_pipelines.feature_extractors.bat.Model(threshold=0.5, **kwargs)[source]

Bases: ModelBaseClass

classifier_predictions(cls_token)[source]
preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.batdetect2_clip_avg module

class bacpipe.model_pipelines.feature_extractors.batdetect2_clip_avg.Model(segment_duration=1, **kwargs)[source]

Bases: ModelBaseClass

classifier_predictions(embeddings)[source]
preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.batdetect2_dets_avg module

class bacpipe.model_pipelines.feature_extractors.batdetect2_dets_avg.Model(segment_duration=1, detection_threshold=0.3, top_k_detections=None, **kwargs)[source]

Bases: ModelBaseClass

preprocess(audio)[source]
bacpipe.model_pipelines.feature_extractors.batdetect2_dets_avg.get_mean_detection_features(results, features, top_k=None)[source]
Return type:

tuple[Tensor, Tensor]

bacpipe.model_pipelines.feature_extractors.beats module

class bacpipe.model_pipelines.feature_extractors.beats.BeatsModel(checkpoint_path)[source]

Bases: object

get_embeddings(spectrogram_input)[source]

Taken from the BEATS forward call. Adapted to work based on the spectrogram input to enable visualization of spectrograms for model result interpretation.

Parameters:

spectrogram_input (torch.Tensor) – batched spectrograms from self.model.preprocess

Returns:

batched embeddings

Return type:

torch.Tensor

class bacpipe.model_pipelines.feature_extractors.beats.Model(**kwargs)[source]

Bases: ModelBaseClass

preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.biolingual module

class bacpipe.model_pipelines.feature_extractors.biolingual.Model(**kwargs)[source]

Bases: ModelBaseClass

preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.birdaves_especies module

class bacpipe.model_pipelines.feature_extractors.birdaves_especies.Model(**kwargs)[source]

Bases: Model

bacpipe.model_pipelines.feature_extractors.birdmae module

class bacpipe.model_pipelines.feature_extractors.birdmae.Model(**kwargs)[source]

Bases: ModelBaseClass

preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.birdnet module

class bacpipe.model_pipelines.feature_extractors.birdnet.Model(**kwargs)[source]

Bases: ModelBaseClass

classifier_predictions(embeddings)[source]
preprocess(audio)[source]
class bacpipe.model_pipelines.feature_extractors.birdnet.Rebuilder(model)[source]

Bases: object

build_model(model)[source]
rebuild_layer(layer)[source]

bacpipe.model_pipelines.feature_extractors.convnext_birdset module

class bacpipe.model_pipelines.feature_extractors.convnext_birdset.Model(**kwargs)[source]

Bases: ModelBaseClass

classifier_predictions(embeddings)[source]
preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.google_whale module

class bacpipe.model_pipelines.feature_extractors.google_whale.Model(**kwargs)[source]

Bases: Model

classifier_predictions(embeddings)[source]

bacpipe.model_pipelines.feature_extractors.hbdet module

class bacpipe.model_pipelines.feature_extractors.hbdet.Model(**kwargs)[source]

Bases: ModelBaseClass

preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.insect459 module

class bacpipe.model_pipelines.feature_extractors.insect459.Model(**kwargs)[source]

Bases: ModelBaseClass

preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.insect66 module

class bacpipe.model_pipelines.feature_extractors.insect66.Model(**kwargs)[source]

Bases: ModelBaseClass

preprocess(audio)[source]
class bacpipe.model_pipelines.feature_extractors.insect66.SpectrogramCNN(cfg, init_backbone=True)[source]

Bases: Module

__init__(cfg, init_backbone=True)[source]

Pytorch network class containing the transformation from waveform to mel spectrogram, as well as the forward pass through a CNN backbone.

Data augmentation like mixup or masked frequency or time can also be applied here.

Parameters:

  • cfg (SimpleNameSpace containing all configurations)

  • init_backbone (bool (Default=True). Whether to download and initialize the backbone.) – Not always necessary when debugging.

bacpipe.model_pipelines.feature_extractors.mix2 module

class bacpipe.model_pipelines.feature_extractors.mix2.Model(**kwargs)[source]

Bases: ModelBaseClass

preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.naturebeats module

class bacpipe.model_pipelines.feature_extractors.naturebeats.Model(**kwargs)[source]

Bases: ModelBaseClass

preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.perch_bird module

class bacpipe.model_pipelines.feature_extractors.perch_bird.Model(**kwargs)[source]

Bases: Model

bacpipe.model_pipelines.feature_extractors.perch_v2 module

class bacpipe.model_pipelines.feature_extractors.perch_v2.Model(model_choice='perch_v2_cpu', sr=32000, segment_length=160000, **kwargs)[source]

Bases: ModelBaseClass

classifier_predictions(embeddings)[source]
preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.protoclr module

class bacpipe.model_pipelines.feature_extractors.protoclr.Model(**kwargs)[source]

Bases: ModelBaseClass

preprocess(audio)[source]
class bacpipe.model_pipelines.feature_extractors.protoclr.Normalization[source]

Bases: Module

forward(x)[source]

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

bacpipe.model_pipelines.feature_extractors.rcl_fs_bsed module

class bacpipe.model_pipelines.feature_extractors.rcl_fs_bsed.Model(**kwargs)[source]

Bases: ModelBaseClass

preprocess(audio)[source]

bacpipe.model_pipelines.feature_extractors.surfperch module

class bacpipe.model_pipelines.feature_extractors.surfperch.Model(**kwargs)[source]

Bases: Model

bacpipe.model_pipelines.feature_extractors.vggish module

class bacpipe.model_pipelines.feature_extractors.vggish.Model(**kwargs)[source]

Bases: Model

Module contents