Datasets:

imageomics
/

upper-waiakea-PAM

license: cc-by-4.0
language:
  - en
pretty_name: Upper Waiākea Forest Reserve Passive Recordings
task_categories:
  - audio-classification
modalities:
  - Audio
tags:
  - biology
  - audio
  - bioacoustics
  - animals
  - CV
  - Hawaiian birds
  - kipuka
  - fragmented habitat
  - unlabeled
  - passive acoustic monitoring
size_categories: 1K<n<10K
description: >-
  This is a dataset containing unlabelled, unprocessed passive acoustic
  recordings of Hawaiian birds in the Upper Waiākea Forst Reserve in Hawaii.
  This dataset is intended for use in unsupervised audio analysis methods,
  classification using existing models, and other machine learning and ecology
  research purposes.

Dataset Card for Upper Waiākea Forest Reserve Passive Recordings

Dataset Details

Dataset Description

Curated by: Namrata Banerji, Ekaterina Nepovinnykh, Jacob Beattie, Hikaru Keebler, Elizabeth Campolongo, Leonardo Teixeira Viotti, Amanda Navine, Patrick Hart, Tanya Berger-Wolf, and Kaiya Provost
Repository: [https://github.com/Imageomics/amakiki-project]

Dataset Description

This is a dataset containing unlabelled, unprocessed passive acoustic recordings of Hawaiian birds in the Upper Waiākea Forst Reserve in Hawaii. This dataset is intended for use in unsupervised audio analysis methods, classification using existing models, and other machine learning and ecology research purposes.

Supported Tasks and Leaderboards

This dataset contains passive acoustic recordings collected as part of the Fall 2024/Spring 2025 Experiential Introduction to AI and Ecology course through the Imageomics Institute and ABC Global Center during January 2025.

This dataset is intended for use with unsupervised computer vision or acoustic machine learning models. No labels are provided, but recorder locations and recording timestamps are included, allowing for analysis of the relationship between ecological factors and variations in birdsong.

The dataset contains ~1623 hours of recording from 19 different recorders located in the Upper Waiākea Forest Reserve.

Dataset Structure

/upper-waiakea-PAM/
    <recorder_id>/
          <recorder_id>_Summary.txt
          Data/
              <recorder_id>_YYYYMMDD_HHMMSS.wav
              ...
    <recorder_id>/
        ...
    ...
    kipuka_metadata.csv
    check.py
    README.md

Data Instances

All audio files are named (recorder_id)-YYYYMMDD-HHMMSS.wav inside a folder named after the recorder id. Each recording starts at the time listed in the filename. Most recordings are 1 hour long, but some may be shorter. Recordings were taken using a SongMeter Micro 2.

kipuka_metadata.csv provides metadata for each recorder, including latitude/longitude and deployment/retrieval dates.

Data Fields

recorder_id, card_code, point_id, deployment_date, retrieval_date, latitude, longitude

kipuka_metadata.csv

recorder_id: Unique identifier for each recorder. Corresponds to the manufacturer ID found on each SongMeter recorder used.
card_code: Unique identifier for SD card used in each recorder
point_id: Unique identifier for each point where a recorder was placed
deployment_date: Date the recorder was deployed
retrieval_date: Date the recorder was retrieved.
latitude: Latitude of recorder
longitude: Longitude of recorder

Data Splits

Only one data split: data. If being used for training/testing/validation of models, splits must be made manually.

Running `check.py`

check.py is a script that can be run to provide a summary of the data for each recorder, such as number of recordings and total recording length. check.py outputs the summary to recorder_data_summary.txt. check.py can be run using the command python check.py with most python environments. If you are having difficulty running this script, try updating your python version, as it has been validated to work using python version 3.12.

The output will be similar to the following:

Summary generated on: 2025-02-14 11:35:04
Folder, File Count, Total Size (MB), <59min Count, >=59min Count, Total Duration (hour)
2MM00549, 95, 15161.5546, 4, 91, 92.00
2MM01345, 83, 13184.8798, 4, 79, 80.01
2MM01340, 88, 14078.0845, 4, 84, 85.43
2MM01695, 87, 14055.8028, 3, 84, 85.29
2MM00813, 80, 12975.0388, 3, 77, 78.73
2MM01690, 85, 13734.8250, 3, 82, 83.34
2MM01356, 87, 14068.1153, 3, 84, 85.37
2MM01186, 86, 13741.1259, 3, 83, 83.38
2MM01792, 87, 14061.7154, 3, 84, 85.33
2MM01707, 99, 15928.7895, 3, 96, 96.66
2MM00926, 88, 14044.4240, 3, 85, 85.22
2MM01028, 82, 12835.9001, 5, 77, 77.89
2MM01088, 88, 14064.3990, 4, 84, 85.34
2MM01007, 79, 12666.8419, 3, 76, 76.86
2MM01631, 87, 14077.6402, 3, 84, 85.42
2MM00471, 86, 13772.1017, 3, 83, 83.57
2MM01323, 99, 15589.7840, 5, 94, 94.60
2MM01655, 97, 15451.0132, 4, 93, 93.76
2MM01339, 89, 14100.2859, 4, 85, 85.56

Dataset Creation

This dataset was compiled as part of the field component of the Experiential Introduction to AI and Ecology Course run by the Imageomics Institute and the AI and Biodiversity Change (ABC) Global Center. This field work was done on the island of Hawai'i January 15-30, 2025. Audio was recorded in the Upper Waiākea Forst Reserve in Hawaii, at the following sites:

Curation Rationale

This dataset was created in order to study Hawaiian bird call variation across kipuka. Passive acoustic monitoring was done to capture Hawaiian bird calls across varying kipuka.

Source Data

These data were originally created by placing recorders in kipuka in the Upper Waiākea Forest Reserve on Hawaii island, recording bird calls.

Data Collection and Processing

Recorder locations were selected to replicate recorder locations from previous work by Sebastián-González, E. and Hart, P.J. (2017). These data have significant overlap the previous recorder locations, while retaining only minimally sufficient overlap with recently collected data to allow for calibration between datasets.

Who are the source data producers?

These data are produced by members of the ABC Global Center and Imageomics Institute, with data collection led by Patrick Hart, Leonardo Viotti, Namrata Banerji, Ekaterina Vepovinnkh, Hikaru Keebler, and Tanya Berger-Wolf, and assistance from all individuals in the AI for Ecology Course.

Considerations for Using the Data

Bias, Risks, and Limitations

These data are unlabelled, unprocessed, and may still contain significant noise due to some recorder's proximity to the road or footpaths. Because of this, humans, cars, or helicopters may also be audible in some recordings.

Additionally, the number of calls recorded for each species is likely long-tailed. Below is a chart depicting the number of occurrences of each species found during different portions of the day using source separation + Perch for species identification:

Recommendations

Consider the impact that raw, unprocessed data may have on use cases for these data. Employing source separation or audio preprocessing methods may be beneficial to downstream analyses.

Licensing Information

This dataset is available to share and adapt for any use under the CC BY 4.0 license, provided appropriate credit is given. We ask that you cite this dataset if you make use of these data in any work or product.

Citation

@misc{
  upper_waiakea_pam,
  title = {Upper Waiākea Forest Reserve Passive Recordings},
  year = {2025},
  url = {https://huggingface.co/datasets/imageomics/upper-waiakea-PAM},
  author = {Banerji, Namrata and Nepovinnykh, Ekaterina and Beattie, Jacob and Keebler, Hikaru and Campolongo, Elizabeth and Teixeira Viotti, Leonardo and Navine, Amanda and Hart, Patrick and Berger-Wolf, Tanya and Provost, Kaiya},
  doi = {coming soon}
  publisher = {Hugging Face}
}

BibTeX:

Acknowledgements

This work was supported by both the Imageomics Institute and the AI and Biodiversity Change (ABC) Global Center. The Imageomics Institute is funded by the US National Science Foundation's Harnessing the Data Revolution (HDR) program under Award #2118240 (Imageomics: A New Frontier of Biological Information Powered by Knowledge-Guided Machine Learning). The ABC Global Center is funded by the US National Science Foundation under Award No. 2330423 and Natural Sciences and Engineering Research Council of Canada under Award No. 585136. Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation or Natural Sciences and Engineering Research Council of Canada.

This material is based in part upon work supported by the National Ecological Observatory Network (NEON), a program sponsored by the U.S. National Science Foundation (NSF) and operated under cooperative agreement by Battelle.

Additionally, we would like to thank Patrick Hart from UH Hilo, Ben Gottesman and Aaron Rice from the Cornell Lab of Ornithology, Mike Long, Shea Uehana, Eissas Ouk, Evan Donoso, Avery Dean, and Ann Carey from the National Ecological Observatory Network (NEON), the UH Hilo LOHE Lab, and the Upper Waiākea Forst Reserve in Hawaii.

Dataset Card Authors

Namrata Banerji, Jacob Beattie, Hikaru Keebler, Kate Nepovinnykh, and Elizabeth Campolongo

Dataset Card Contact

banerji.8@osu.edu beattie.74@osu.edu nepove@rpi.edu keebler.8@osu.edu

Dataset Card for Upper Waiākea Forest Reserve Passive Recordings

Dataset Details

Dataset Description

Dataset Description

Supported Tasks and Leaderboards

Dataset Structure

Data Instances

Data Fields

Data Splits

Running check.py

Dataset Creation

Curation Rationale

Source Data

Data Collection and Processing

Who are the source data producers?

Considerations for Using the Data

Bias, Risks, and Limitations

Recommendations

Licensing Information

Citation

Acknowledgements

Dataset Card Authors

Dataset Card Contact

Running `check.py`