Fix augmentations

batdetect2/train/__init__.py

@@ -18,7 +18,6 @@ from batdetect2.train.config import TrainingConfig, load_train_config
 from batdetect2.train.dataset import (
     LabeledDataset,
     RandomExampleSource,
-    SubclipConfig,
     TrainExample,
     list_preprocessed_files,
 )
@@ -37,7 +36,6 @@ __all__ = [
     "LabeledDataset",
     "LossFunction",
     "RandomExampleSource",
-    "SubclipConfig",
     "TimeMaskAugmentationConfig",
     "TrainExample",
     "TrainerConfig",

batdetect2/train/augmentations.py

@@ -126,7 +126,8 @@ def mix_examples(
     audio1 = example["audio"]
     audio2 = adjust_width(other["audio"].values, len(audio1))
 
-    combined = weight * audio1 + (1 - weight) * audio2
+    with xr.set_options(keep_attrs=True):
+        combined = weight * audio1 + (1 - weight) * audio2
 
     spectrogram = preprocessor.compute_spectrogram(
         combined.rename({"audio_time": "time"})
@@ -227,7 +228,6 @@ def add_echo(
         'spectrogram' variable recomputed. Other variables (targets, attrs)
         are copied from the original example.
     """
-
     if delay is None:
         delay = np.random.uniform(0, max_delay)
 
@@ -237,7 +237,9 @@ def add_echo(
     audio = example["audio"]
     step = arrays.get_dim_step(audio, "audio_time")
     audio_delay = audio.shift(audio_time=int(delay / step), fill_value=0)
-    audio = audio + weight * audio_delay
+
+    with xr.set_options(keep_attrs=True):
+        audio = audio + weight * audio_delay
 
     spectrogram = preprocessor.compute_spectrogram(
         audio.rename({"audio_time": "time"}),
@@ -260,6 +262,7 @@ def add_echo(
             data=spectrogram,
             dims=example["spectrogram"].dims,
             coords=example["spectrogram"].coords,
+            attrs=example["spectrogram"].attrs,
         ),
     )
 
@@ -311,7 +314,8 @@ def scale_volume(
     if factor is None:
         factor = np.random.uniform(min_scaling, max_scaling)
 
-    return example.assign(spectrogram=example["spectrogram"] * factor)
+    with xr.set_options(keep_attrs=True):
+        return example.assign(spectrogram=example["spectrogram"] * factor)
 
 
 class WarpAugmentationConfig(BaseConfig):
@@ -366,51 +370,52 @@ def warp_spectrogram(
         example.time.size,
     )
 
-    spectrogram = (
-        example["spectrogram"]
-        .interp(
-            coords={"time": new_time},
-            method="linear",
+    with xr.set_options(keep_attrs=True):
+        spectrogram = (
+            example["spectrogram"]
+            .interp(
+                coords={"time": new_time},
+                method="linear",
+                kwargs=dict(
+                    fill_value=0,
+                ),
+            )
+            .clip(min=0)
+        )
+
+        detection = example["detection"].interp(
+            time=new_time,
+            method="nearest",
             kwargs=dict(
                 fill_value=0,
             ),
         )
-        .clip(min=0)
-    )
 
-    detection = example["detection"].interp(
-        time=new_time,
-        method="nearest",
-        kwargs=dict(
-            fill_value=0,
-        ),
-    )
+        classification = example["class"].interp(
+            time=new_time,
+            method="nearest",
+            kwargs=dict(
+                fill_value=0,
+            ),
+        )
 
-    classification = example["class"].interp(
-        time=new_time,
-        method="nearest",
-        kwargs=dict(
-            fill_value=0,
-        ),
-    )
+        size = example["size"].interp(
+            time=new_time,
+            method="nearest",
+            kwargs=dict(
+                fill_value=0,
+            ),
+        )
 
-    size = example["size"].interp(
-        time=new_time,
-        method="nearest",
-        kwargs=dict(
-            fill_value=0,
-        ),
-    )
-
-    return example.assign(
-        {
-            "time": new_time,
-            "spectrogram": spectrogram,
-            "detection": detection,
-            "class": classification,
-            "size": size,
-        }
-    )
+        return example.assign(
+            {
+                "time": new_time,
+                "spectrogram": spectrogram,
+                "detection": detection,
+                "class": classification,
+                "size": size,
+            }
+        )
 
 
 def mask_axis(
@@ -423,7 +428,7 @@ def mask_axis(
     """Mask values along a specified dimension.
 
     Sets values in the DataArray to `mask_value` where the coordinate along
-    `dim` falls within the range [`start`, `end`). Values outside this range
+    `dim` falls within the range (`start`, `end`). Values outside this range
     are kept. Used as a helper for time/frequency masking.
 
     Parameters
@@ -462,7 +467,8 @@ def mask_axis(
     if callable(mask_value):
         mask_value = mask_value(array)
 
-    return array.where(condition, other=mask_value)
+    with xr.set_options(keep_attrs=True):
+        return array.where(condition, other=mask_value)
 
 
 class TimeMaskAugmentationConfig(BaseConfig):
@@ -511,7 +517,8 @@ def mask_time(
         end = start + mask_size
         spectrogram = mask_axis(spectrogram, "time", start, end)
 
-    return example.assign(spectrogram=spectrogram)
+    with xr.set_options(keep_attrs=True):
+        return example.assign(spectrogram=spectrogram)
 
 
 class FrequencyMaskAugmentationConfig(BaseConfig):
@@ -566,7 +573,8 @@ def mask_frequency(
         end = start + mask_size
         spectrogram = mask_axis(spectrogram, "frequency", start, end)
 
-    return example.assign(spectrogram=spectrogram)
+    with xr.set_options(keep_attrs=True):
+        return example.assign(spectrogram=spectrogram)
 
 
 AugmentationConfig = Annotated[

batdetect2/train/preprocess.py

@@ -26,6 +26,7 @@ from pathlib import Path
 from typing import Callable, Optional, Sequence
 
 import xarray as xr
+from loguru import logger
 from soundevent import data
 from tqdm.auto import tqdm