FEAT: exposed max_segment_length as config param + black

CONFIG.yaml

@@ -21,6 +21,7 @@ data:
   overlap: 0.5 # used in preprocessing
   n_subsample: 1 
   num_mel_bins: 128 # used in preprocessing
+  max_segment_length: 1.0 # used in preprocessing
   status: train # used in preprocessing, train or validate or evaluate
 
 

data_utils/DCASEfewshot.py

@@ -33,7 +33,6 @@
 PLOT = False
 PLOT_TOO_SHORT_SAMPLES = False
 PLOT_SUPPORT = False
-MAX_SEGMENT_LENGTH = 1.0
 
 
 def normalize_mono(samples):
@@ -95,6 +94,7 @@ def prepare_training_val_data(
     target_fs=16000,
     overlap=0.5,
     num_mel_bins=128,
+    max_segment_length=1.0,
 ):
     """Prepare the Training_Set
 
@@ -120,7 +120,7 @@ def preprocess_df(df):
                 continue
             # obtain a segment with large margins around event
             extra_time = 3
-            segment_length_here = min(min_segment_lengths[label], MAX_SEGMENT_LENGTH)
+            segment_length_here = min(min_segment_lengths[label], max_segment_length)
             frame_shift = np.round(segment_length_here / tensor_length * 1000)
             frame_shift = 1 if frame_shift < 1 else frame_shift
             start_waveform = int((df["Starttime"][ind] - extra_time) * fs)
@@ -152,7 +152,7 @@ def preprocess_df(df):
                 sample_frequency=target_fs,
                 frame_length=frame_length,
                 frame_shift=frame_shift,
-                num_mel_bins=num_mel_bins
+                num_mel_bins=num_mel_bins,
             )
             data = fbank.data[0].T
             # select the relevant segment (without the large margins)
@@ -262,7 +262,8 @@ def preprocess_df(df):
         "tensor_length": tensor_length,
         "set_type": set_type,
         "overlap": overlap,
-        "num_mel_bins": num_mel_bins
+        "num_mel_bins": num_mel_bins,
+        "max_segment_length": max_segment_length,
     }
     if resample:
         my_hash_dict["target_fs"] = target_fs
@@ -282,7 +283,7 @@ def preprocess_df(df):
         os.makedirs(os.path.join(target_path, "plots"))
 
     # Save my_hash_dict as a metadata file
-    with open(os.path.join(target_path, 'metadata.json'), 'w') as f:
+    with open(os.path.join(target_path, "metadata.json"), "w") as f:
         json.dump(my_hash_dict, f)
 
     print("=== Processing data ===")
@@ -313,9 +314,9 @@ def preprocess_df(df):
         audio_path = file.replace("csv", "wav")
         print("Processing file name {}".format(audio_path))
         y, fs = librosa.load(audio_path, sr=None, mono=True)
-        if not resample or my_hash_dict["target_fs"]> fs:
+        if not resample or my_hash_dict["target_fs"] > fs:
             target_fs = fs
-        else: 
+        else:
             target_fs = my_hash_dict["target_fs"]
         df = df[(df == "POS").any(axis=1)]
         df = df.reset_index()
@@ -372,7 +373,7 @@ def preprocess_df(df):
             # CREATE QUERY SETS
             # obtain file specific frame_shift and save to meta.csv
 
-            segment_length_here = min(min_segment_lengths["POS"], MAX_SEGMENT_LENGTH)
+            segment_length_here = min(min_segment_lengths["POS"], max_segment_length)
             frame_shift = np.round(segment_length_here / tensor_length * 1000)
             frame_shift = 1 if frame_shift < 1 else frame_shift
             print(file_name)
@@ -566,7 +567,6 @@ def preprocess_df(df):
         type=str,
     )
 
-
     # get input
     cli_args = parser.parse_args()
 
@@ -578,19 +578,28 @@ def preprocess_df(df):
         cfg = yaml.load(f, Loader=FullLoader)
 
     # Check values in config file
-    if not (cfg["data"]["status"]=="train" or cfg["data"]["status"]=="validate" or cfg["data"]["status"]=="test"):
-        raise Exception("ERROR: "+ str(cli_args.config) + ": Accepted values for 'status' are 'train', 'validate', or 'test'. Received '" + str(cfg["data"]["status"]) + "'.")
-
+    if not (
+        cfg["data"]["status"] == "train"
+        or cfg["data"]["status"] == "validate"
+        or cfg["data"]["status"] == "test"
+    ):
+        raise Exception(
+            "ERROR: "
+            + str(cli_args.config)
+            + ": Accepted values for 'status' are 'train', 'validate', or 'test'. Received '"
+            + str(cfg["data"]["status"])
+            + "'."
+        )
+
     # Select 'set_type' depending on chosen status
-    if cfg["data"]["status"]=="train":
+    if cfg["data"]["status"] == "train":
         cfg["data"]["set_type"] = "Training_Set"
 
-    elif cfg["data"]["status"]=="validate":
+    elif cfg["data"]["status"] == "validate":
         cfg["data"]["set_type"] = "Validation_Set"
 
     else:
         cfg["data"]["set_type"] = "Evaluation_Set"
-
 
     prepare_training_val_data(
         cfg["data"]["status"],
@@ -604,4 +613,5 @@ def preprocess_df(df):
         cfg["data"]["target_fs"],
         cfg["data"]["overlap"],
         cfg["data"]["num_mel_bins"],
+        cfg["data"]["max_segment_length"],
     )

datamodules/DCASEDataModule.py

@@ -103,8 +103,7 @@ def __init__(
         n_subsample: int = 1,
         overlap: float = 0.5,
         num_mel_bins: int = 128,
-
-        **kwargs,
+        max_segment_length: float = 1.0**kwargs,
     ):
         super().__init__(**kwargs)
         self.n_task_train = n_task_train
@@ -123,6 +122,7 @@ def __init__(
         self.n_subsample = n_subsample
         self.overlap = overlap
         self.num_mel_bins = num_mel_bins
+        self.max_segment_length = max_segment_length
         self.setup()
 
     def setup(self, stage=None):
@@ -136,6 +136,7 @@ def setup(self, stage=None):
             "set_type": self.set_type,
             "overlap": self.overlap,
             "num_mel_bins": self.num_mel_bins,
+            "max_segment_length": self.max_segment_length,
         }
         if self.resample:
             my_hash_dict["target_fs"] = self.target_fs

evaluate/evaluateDCASE.py

@@ -75,10 +75,11 @@ def train_model(
         try:
             pretrained_model = ProtoBEATsModel.load_from_checkpoint(pretrained_model)
         except KeyError:
-            print("Failed to load the pretrained model. Please check the checkpoint file.")
+            print(
+                "Failed to load the pretrained model. Please check the checkpoint file."
+            )
             return None
 
-
     # train the model
     trainer.fit(model, datamodule=datamodule_class)
 
@@ -296,7 +297,11 @@ def main(
     # Train the model with the support data
     print("[INFO] TRAINING THE MODEL FOR {}".format(filename))
 
-    model = training(cfg["model"]["model_path"], custom_dcasedatamodule, max_epoch=cfg["trainer"]["max_epochs"])
+    model = training(
+        cfg["model"]["model_path"],
+        custom_dcasedatamodule,
+        max_epoch=cfg["trainer"]["max_epochs"],
+    )
 
     # Get the prototypes coordinates
     a = custom_dcasedatamodule.test_dataloader()
@@ -364,7 +369,9 @@ def main(
         # Train the model with the support data
         print("[INFO] TRAINING THE MODEL FOR {}".format(filename))
 
-        model = training(cfg["model"]["model_path"], custom_dcasedatamodule, max_epoch=1)
+        model = training(
+            cfg["model"]["model_path"], custom_dcasedatamodule, max_epoch=1
+        )
 
         # Get the prototypes coordinates
         a = custom_dcasedatamodule.test_dataloader()
@@ -424,7 +431,9 @@ def main(
     )
 
     result_POS_merged = merge_preds(
-        df=result_POS, tolerence=cfg["tolerance"], tensor_length=cfg["data"]["tensor_length"]
+        df=result_POS,
+        tolerence=cfg["tolerance"],
+        tensor_length=cfg["data"]["tensor_length"],
     )
 
     # Add the filename
@@ -478,19 +487,43 @@ def write_wav(
     # Expand the dimensions
     gt_labels = np.repeat(
         np.squeeze(gt_labels, axis=1).T,
-        int(cfg["data"]["tensor_length"] * cfg["data"]["overlap"] * target_fs * frame_shift / 1000),
+        int(
+            cfg["data"]["tensor_length"]
+            * cfg["data"]["overlap"]
+            * target_fs
+            * frame_shift
+            / 1000
+        ),
     )
     pred_labels = np.repeat(
         pred_labels.T,
-        int(cfg["data"]["tensor_length"] * cfg["data"]["overlap"] * target_fs * frame_shift / 1000),
+        int(
+            cfg["data"]["tensor_length"]
+            * cfg["data"]["overlap"]
+            * target_fs
+            * frame_shift
+            / 1000
+        ),
     )
     distances_to_pos = np.repeat(
         distances_to_pos.T,
-        int(cfg["data"]["tensor_length"] * cfg["data"]["overlap"] * target_fs * frame_shift / 1000),
+        int(
+            cfg["data"]["tensor_length"]
+            * cfg["data"]["overlap"]
+            * target_fs
+            * frame_shift
+            / 1000
+        ),
     )
     z_scores_pos = np.repeat(
         z_scores_pos.T,
-        int(cfg["data"]["tensor_length"] * cfg["data"]["overlap"] * target_fs * frame_shift / 1000),
+        int(
+            cfg["data"]["tensor_length"]
+            * cfg["data"]["overlap"]
+            * target_fs
+            * frame_shift
+            / 1000
+        ),
     )
 
     # pad with zeros
@@ -575,10 +608,10 @@ def write_wav(
     version_name = os.path.basename(version_path)
 
     # Select 'set_type' depending on chosen status
-    if cfg["data"]["status"]=="train":
+    if cfg["data"]["status"] == "train":
         cfg["data"]["set_type"] = "Training_Set"
 
-    elif cfg["data"]["status"]=="validate":
+    elif cfg["data"]["status"] == "validate":
         cfg["data"]["set_type"] = "Validation_Set"
 
     else:
@@ -593,7 +626,8 @@ def write_wav(
         "tensor_length": cfg["data"]["tensor_length"],
         "set_type": cfg["data"]["set_type"],
         "overlap": cfg["data"]["overlap"],
-        "num_mel_bins": cfg["data"]["num_mel_bins"]
+        "num_mel_bins": cfg["data"]["num_mel_bins"],
+        "max_segment_length": cfg["data"]["max_segment_length"],
     }
     if cfg["data"]["resample"]:
         my_hash_dict["target_fs"] = cfg["data"]["target_fs"]
@@ -617,11 +651,11 @@ def write_wav(
         "support_labels_*.npy",
     )
     query_data_path = os.path.join(
-        "/data/DCASEfewshot", 
-        cfg["data"]["status"], 
-        hash_dir_name, 
-        "audio", 
-        "query_data_*.npz"
+        "/data/DCASEfewshot",
+        cfg["data"]["status"],
+        hash_dir_name,
+        "audio",
+        "query_data_*.npz",
     )
     query_labels_path = os.path.join(
         "/data/DCASEfewshot",
@@ -631,11 +665,7 @@ def write_wav(
         "query_labels_*.npy",
     )
     meta_df_path = os.path.join(
-        "/data/DCASEfewshot", 
-        cfg["data"]["status"], 
-        hash_dir_name, 
-        "audio", 
-        "meta.csv"
+        "/data/DCASEfewshot", cfg["data"]["status"], hash_dir_name, "audio", "meta.csv"
     )
 
     # set target path