{
  "fetched_at": "2026-10-19T07:17:15Z",
  "source_url": "https://www.kaggle.com/competitions/datafestintegration2023/data",
  "pages": [
    {
      "name": "data-description",
      "content": "The **dataset is annotated by the DataFest Team at Indian Statistical Institute Kolkata**. The occurrences of A calls were identified and labelled from underwater audio recordings captured by deep-sea hydrophone devices. The recording spanned almost a month. \n\nThe number of samples per class:\n- **No A calls (Labeled as 0)**: 12952\n- **A calls (Labeled as 1)**: 12996\n- **Unlabelled (test)**: 2000\n\n*Visualization of an audio file from the training dataset*\n| Waveform | Spectogram |\n| --- | --- |\n| ![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6668389%2F4385645263adc06e4be20011217a77d1%2FScreenshot%202023-01-16%20at%202.54.43%20AM.png?generation=1673817906901752&alt=media) | ![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6668389%2F8471434d8d6fb5bc0958d2db6bd788ac%2FScreenshot%202023-01-16%20at%202.51.32%20AM.png?generation=1673817800473520&alt=media) |\n\n### 📈Updates\n\n- Data was last updated on 15/01/2023 at 5:00 PM IST\n- Update Logs:\n  - The quality of data is enhanced.\n  - Kindly use the updated data. It is final and hopefully, no further changes would be made.\n\n## 📂 Files\n\nThis is a brief description of the folders and files present:\n*   **train** - the training set folder consisting of subfolders **1** ('A' calls) and **0** (No 'A' calls). Each of these subfolders contains audio clips (.wav).\n*   **test** - the test set folder consisting of audio clips (.wav) on which prediction has to be generated.\n*   **sample_submission.csv** - a sample submission file in the correct format for the audio clips in the *test* folder\n\nNote: No **train.csv** is provided as the **train** directory structure has the labels mentioned as its subfolder names i.e. 0 and 1.\n\n##  🔍 Submission Columns\n\nYour submission `.csv` file should contain the following columns:\n*   `id` - audio clip id\n*   `label` - predicted label\n\nCheck the `sample_submission.csv` for the format. Predict the labels for all the audio clips present in the `test` folder and put them in your `.csv` submission file. \n\n⚠️ **Cautionary Note**: To err is human. The dataset labels are not completely error free as most of it is annotated by human annotators and some mislabelling is also present. ",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "## 🚀 Introduction\n\n**☀️ [Bonus Problem](https://www.kaggle.com/competitions/im-hard-to-spot) is LIVE NOW!!**\n\n🌊 The deep blue sea is home to a plethora of fascinating and mysterious creatures. Sound is the fundamental way these giants communicate with each other, but much of their communication remains a mystery to scientists. \n\n🧩 Scientists have been fascinated by these humongous mammals for many decades, and their vocalizations have been a particularly intriguing area of study. Yes, You got it right! We are talking about Blue whales 🐋. \n\n🎶 They are highly intelligent and social species who are known for their low-frequency, repetitive \"songs\" that can travel for thousands of miles through the ocean. These songs are thought to be used for communication between individual whales, and they can change over time, with different populations singing different songs.\n\n🔬 Studying the vocalizations of blue whales can provide valuable insights into the behaviour and communication patterns of these animals. This knowledge can also aid conservation efforts by helping scientists track and monitor blue whale populations.\n\n🎛 However, researching blue whales' vocalization is not an easy task, and requires specialized equipment and expertise to collect and analyze the audio data. Additionally, blue whales are a rare and elusive species, making it challenging to obtain a large enough dataset to train accurate models.\n\n🎯 In this competition, we invite you to help unlock the secrets of blue whale communication by **building models that can accurately identify various calls made by these creatures**. Your participation will not only be an exciting challenge, but also a valuable contribution to the field of marine biology and conservation efforts. So, put on your thinking caps, sharpen your machine learning skills and let's work together to understand these magnificent creatures better. Good luck!\n\n➡️ **[REGISTER HERE](https://forms.gle/fyXABdTqATa1n52L6)** **(Deadline: 30th January 2023 - Extended)**\nℹ️ **Participants are requested to go through the [Rules](https://www.kaggle.com/competitions/datafestintegration2023/rules) carefully.**\n⬇️ **Participants are requested to go through each of the informative sections below properly.**\n\n## 🐳 Whale Vocalizations \n\n![An example of whale call spectrogram, for depiction purposes only](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F6668389%2Fe3818556da3bf2761196ba91ea08ff5f%2FScreenshot%202023-01-15%20at%2012.56.00%20PM.png?generation=1673767655080247&alt=media)\n*Visualization of whale songs*\n\nThere are various types of whale vocalizations. The different types of vocalizations and their purposes are still not well understood, and ongoing research continues to uncover new insights into the vocal behaviour of these fascinating animals. Here we focus on a specific vocalization,\n\n🐋 **A Calls**: They are characterized by a low-frequency, ***repetitive pattern of pulses*** that are typically around ***70-90 Hz in frequency***. These calls are known to be used for long-distance communication between individual whales and can be heard over large distances. They are typically produced by adult males and can last for several minutes. \n\nRead more on other types of Whale vocalizations at the links given under the Important Links Section.\n\n## 💡 Helper Code & Hints \n\nWe know many of you might not have worked with sound data before and we are there to provide the necessary help for you to make progress. \n\n🧗🏻 **Browse through the important links below for getting warmed up.**\n\n---\n\n🚧 We will be uploading helper codes on [Code section](https://www.kaggle.com/competitions/datafestintegration2023/code).\n\n🔆 **[HELPER CODE 01](https://www.kaggle.com/code/rishideychowdhury/helper-code-01)**: Working with Audio Data, Visualizing and Listening up.\n\n---\n👀 **Keep checking here for regular updates on hints.**\n\n🛠 **HINT 0**: To make a prediction on whether an A call is present or not, you need to know **What is an A call?** **What is the duration of an A call and the frequency range of an A call?** (A brief description can be found in the above discussion and in the important links below)\n\n🛠 **HINT 1**: Spectrograms are very useful for audio data. **How does an A call differ from a Non-A call?** Can we use that to our advantage?\n\n## 🌟 Bonus Problem\n\n⏱ **Bonus Problem Release Date: 22nd January, 2023 at 11:00 PM IST**\n\n↗️ Visit the bonus problem page: [I'm Hard to Spot](https://www.kaggle.com/competitions/im-hard-to-spot)\n- Submission to this challenge is not a prerequisite for evaluation of the solution submitted to the bonus problem. \n- Grading of this challenge and the bonus problem would be done independently.\n- A weighted average of the scores obtained on both this challenge and the bonus problem will be taken as the final score, with the weight of the bonus problem strictly greater than this challenge.\n\n\n### 🔗 Important Links\n\n- Inspiration Video for the challenge [Fathom — Official Trailer | Apple TV+](https://youtu.be/Y-Mmnk0LZLU). \n  - It is definitely a must-watch!\n- Wikipedia page on [Whale Vocalizations](https://en.wikipedia.org/wiki/Whale_vocalization). \n  - Great for listening to audio samples of various whales from across the oceans of the world.\n- [Different patterns in Whale Vocalizations](https://www.mbari.org/news/patterns-in-blue-whale-songs-show-whether-the-whales-are-feeding-or-migrating/) observed in deep sea hydrophones. \n  - **Check the video near the end for the sound patterns corresponding to A calls and other whale calls.**\n- Techniques adopted in a previous [Right Whale Detection Challenge](https://www.kaggle.com/competitions/whale-detection-challenge). \n  - **Very useful for getting some initial directions to start working.*",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "This is a **Binary Classification Task**. The evaluation metric for the Leaderboard on this challenge is the **Micro F1-Score**. This would allow us to evaluate your model in terms of its specificity and sensitivity. \nIn addition to it, **all the teams would be required to submit a 1-2 page report and codes** explaining their approach for our judges to decide on the winning teams. Although a higher position on the Leaderboard is a plus, **Novelty and Performance** will be the deciding factor for the winners.\nNotifications through the mail will be sent to the registered participants about the report format and submission procedure.\n\n---\n\nThe Leaderboard's Micro F1 score measures accuracy using the statistics of precision and recall. It is given by:\n\n$$\\text{F1} = \\frac{tp}{tp + \\frac{1}{2}\\cdot (fp + fn)}$$\n\nwhere tp is the true positive, fp is the false positive and fn is the false negative.\n\n## 📋 Submission Format\n\n**For every participant**, the `.csv` submission file should contain two columns: `id` and `label`. \n\n- `id` is the id of the audio file \n- `label` is the predicted label for that audio file.\n\nThe `.csv` file should contain a header and have the following format:\n```\nid,label\n1,1\n```\nCheck out `sample_submission.csv` in the [Data section](https://www.kaggle.com/competitions/datafestintegration2023/data) for more info.\n\n---\n\n## 💯 Scoring\nThe following describes how the final scores would be obtained for each team,\n- A weighted average of scores given by the cohort of professors and industry experts on both submissions, i.e. the Whale Sound Classification Challenge and the Bonus Problem, independently will be used to arrive at the final score given to the teams.\n- The Bonus Problem is more challenging compared to the [Whale Sound Classification](https://www.kaggle.com/competitions/datafestintegration2023) and hence the weightage for the Bonus Problem's score is strictly greater than that of the whale challenge.\n- In case no submission is made to the bonus problem, the score for the bonus problem will be considered to be 0.\n- In case no submission is made to the Whale Sound Classification problem, the score for that will be considered to be 0.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": "## 🧗‍♂️ Teams \n\n- Participants can work in a **team of a maximum of 3 members**. A [Google Form](https://forms.gle/twdpv3rUXiydDFMB6) has been circulated for the **registration of the teams (Solo/Duo/Trio)**.\n- Kindly **use the same Team Name which you used while registering**. It will help us identify your team, as you trod your way up the leaderboard.\n- Solo participants are **NOT** allowed to team up on Kaggle.\n- Duo teams are **NOT** allowed to team up with someone other than their registered team member on Kaggle.\n- Trio teams are **NOT** allowed to team up with someone other than their registered team members on Kaggle.\n\n## 🏆 Eligibility for Participation and Prizes\n\n- This is **open for all competition**. Anyone with a passion for Machine Learning can participate and compete for awesome prizes.\n-  **Registration is mandatory for being eligible for prizes and awards**.\n- Prize Levels:\n  - **Solo** - Participants, who are registered as Solo will compete against all the other Solo participants and is eligible for Solo Prizes and Awards.\n  - **Duo** - Teams, who are registered as Duo will compete against all the other Duo Teams and is eligible for Duo Prizes and Awards.\n  - **Trio** - Teams, who are registered as Trio will compete against all the other Trio Teams and is eligible for Trio Prizes and Awards.\n  - **Two overall runner-ups from the pool of Solo, Duo and Trio will be awarded**.\n- For the prize amounts check the rules poster below.\n- **Submission of code/implementation details and a report is mandatory to be considered for prizes.** \n- The participants of the **Winning Teams will get a chance to pursue a project under Indian Statistical Institute Faculty**. (Interviews would be conducted)\n- Refer to the rules poster below for more detail on the prizes.\n\n*Unfortunately, this year we can not ship Merchandise outside of India.*\n\n## ☘️ Judging\n\n- **Mails will be sent to the registered participants in the last week for the report and code submission. The results of your experiments should be reproducible**.\n- The judgement of winners will be decided by **a team of professors at the Indian Statistical Institute Kolkata and industry experts**. \n- Judging criteria: Though a higher position on the leaderboard is advantageous, the final judgement would be based on **Novelty and Performance** of your model.\n- A bonus problem is provided, submissions to which will be evaluated independently to submissions made to this challenge. **A weighted average of the scores of both the challenge and bonus problem solution will be taken as the final score of a team. (The weight of the bonus problem is higher owing to its greater difficulty).**\n\n## 🕹 Submission Limits\n\n- You may submit a **maximum of 6 entries per day**.\n- You may select **up to 2 final submissions for judging**.\n\n## 🏁 Competition Timeline\n\nThe following are the submission deadlines for submissions to this challenge as well as the bonus problem.\n\n- **Start Date: 15/01/2023 8:00 PM IST**\n- **Entry Deadline: 30/01/2023 (Extended)**\n- **End Date: 05/02/2023 12:00 AM IST**\n\n## ❓ FAQs\n\n1. **Can I use pre-trained models?**\n- **Using pre-trained models (from other sources), which have been trained on large ocean/sea recording or sea animal sounds** is **NOT allowed**.\n- **You are allowed to pre-train your model on larger datasets/sea recordings/sea animal sounds/others on your own.** The relevant training code and training epoch results need to be submitted in the code/implementation detail.\n2. **Are there any restrictions on the tools/data/techniques that can be used for the bonus problem?**\n- There is absolutely no restriction on the things you can try. We encourage you to come up with creative and innovative ways to tackle such a real-world problem with all you have at your disposal.\n3. **How will I know after mailing whether my report/notebook is submitted?**\n- We will reply to you back with a confirmation message upon receiving your submission.\n\nNote:\n\n- The organizers will take the final call on the final prize money as well as any modification of the evaluation criteria (if any).\n- Not abiding by any of the above rules might lead to disqualification.\n\n### Rules Poster\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F6668389%2F61d858b0d16004d4ad695ee9cd141717%2FDataFest-Rules-2023.png?generation=1674360622575743&alt=media)",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Event Poster",
      "content": "| 22/01/2023 | 15/01/2023 | 08/01/2023 |\n| --- | --- | --- |\n|![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F6668389%2F207643c13da7e0247c32f1a0f1516280%2FDataFest-Bonus-Poster.png?generation=1674423367211859&alt=media)  | ![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F6668389%2Fcbb798c5c3f41fe21d6b5c551f9da524%2FDataFest%20Live-Poster.png?generation=1674360552023104&alt=media) |![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F6668389%2F6940fa4205d12d709d89d36b9ef874ec%2FDataFest-2023.png?generation=1674360552050129&alt=media)|",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}