| """run hawkears inference on 14 SBT localization arrays, ~50 recorders each |
| |
| 1.7Tb of audio across all the arrays |
| """ |
|
|
| from bioacoustics_model_zoo import HawkEars_v010 |
| from glob import glob |
| from pathlib import Path |
| from tqdm.autonotebook import tqdm |
|
|
| try: |
| from redacted_paths import REDACTED_OUTPUT_DIR, REDACTED_DATA_ROOT |
| except ImportError: |
| print("redacted paths not available") |
| REDACTED_OUTPUT_DIR = "" |
| REDACTED_DATA_ROOT = "" |
|
|
| data_root = f"{REDACTED_DATA_ROOT}" |
| out_dir = f"{REDACTED_OUTPUT_DIR}/1_detect_hawkears/" |
|
|
| arrays = glob(f"{data_root}/SBT-*") |
| m = HawkEars_v010() |
| m.device = "cuda:0" |
|
|
| print(f"Found {len(arrays)} arrays from SBT dataset") |
| for audio_dir in tqdm(arrays): |
| folders = glob(f"{audio_dir}/SBT*") |
| if len(folders) < 1: |
| continue |
| array_name = Path(audio_dir).name |
| preds_save_dir = f"{out_dir}/{array_name}/" |
| Path(preds_save_dir).mkdir(exist_ok=True) |
|
|
| folders = glob(f"{audio_dir}/SBT*") |
| print(f"Found {len(folders)} audio folders for array {Path(audio_dir).name}") |
|
|
| for f in tqdm(folders): |
| files = glob(f"{f}/*.wav") |
| if len(files) < 1: |
| continue |
| audio_folder_name = Path(f).name |
| |
|
|
| save_dir = f"{preds_save_dir}/{audio_folder_name}" |
| Path(save_dir).mkdir(exist_ok=True) |
|
|
| preds_save_path = f"{save_dir}/{audio_folder_name}_hawkears_preds.csv" |
| if Path(preds_save_path).exists(): |
| |
| continue |
| print(f"running HawkEars prediction on {len(files)} files from {f}") |
|
|
| preds = m.predict(files, num_workers=8, batch_size=256) |
|
|
| |
| preds = preds.reset_index() |
| preds["file"] = preds["file"].apply(lambda x: Path(x).name) |
| |
| preds.to_csv( |
| f"{save_dir}/{Path(f).name}_hawkears_preds.csv", |
| float_format="%.3f", |
| index=False, |
| ) |
|
|
