{
  "id": 316139,
  "title": "Imbalanced Data?",
  "url": "/competitions/happy-whale-and-dolphin/discussion/316139",
  "author_name": "",
  "post_date": "2022-03-31T12:48:21.393946900Z",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hi, I might be too late to ask this question but I'm new to machine learning and was going through the Happywhale Dataset, I realized that many whales and dolphins have only 1 image per individual in the dataset. I was wondering how are people dealing with this imbalance? <br>\nShould I collect more data?<br>\nShould I just copy the same image of the individual multiple times?<br>\nAny answers would be appreciated 😄</p>",
  "messages": [
    {
      "id": "1741006",
      "postDate": "03/31/2022 12:48:21",
      "content": "<p>Hi, I might be too late to ask this question but I'm new to machine learning and was going through the Happywhale Dataset, I realized that many whales and dolphins have only 1 image per individual in the dataset. I was wondering how are people dealing with this imbalance? <br>\nShould I collect more data?<br>\nShould I just copy the same image of the individual multiple times?<br>\nAny answers would be appreciated 😄</p>",
      "rawMarkdown": "Hi, I might be too late to ask this question but I'm new to machine learning and was going through the Happywhale Dataset, I realized that many whales and dolphins have only 1 image per individual in the dataset. I was wondering how are people dealing with this imbalance? \nShould I collect more data?\nShould I just copy the same image of the individual multiple times?\nAny answers would be appreciated 😄",
      "votes": null
    },
    {
      "id": "1743653",
      "postDate": "04/03/2022 07:15:40",
      "content": "<p>The competition context is different that the classical classification or regression problems. Having so many individuals, some with more samples, others with just one  requires a different approach. Using ArcFace the problem is transformed in a resembling recognition more than a clasification. You don't need to worry so much about individual with just one images, focus on image preprocessing, having a good input dataset and post processing techniques on resulting embeddings </p>",
      "rawMarkdown": "The competition context is different that the classical classification or regression problems. Having so many individuals, some with more samples, others with just one  requires a different approach. Using ArcFace the problem is transformed in a resembling recognition more than a clasification. You don't need to worry so much about individual with just one images, focus on image preprocessing, having a good input dataset and post processing techniques on resulting embeddings",
      "votes": null
    },
    {
      "id": "1744634",
      "postDate": "04/04/2022 06:22:25",
      "content": "<p>Thanks, I don't really know what ArcFace is but I've seen many discussions use this term, I will check it out</p>",
      "rawMarkdown": "Thanks, I don't really know what ArcFace is but I've seen many discussions use this term, I will check it out",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1743653,
      "author_name": "vladvdv",
      "author_url": "",
      "post_date": "04/03/2022 07:15:40",
      "content": "<p>The competition context is different that the classical classification or regression problems. Having so many individuals, some with more samples, others with just one  requires a different approach. Using ArcFace the problem is transformed in a resembling recognition more than a clasification. You don't need to worry so much about individual with just one images, focus on image preprocessing, having a good input dataset and post processing techniques on resulting embeddings </p>",
      "votes": null,
      "replies": [
        {
          "id": 1744634,
          "author_name": "wasdac",
          "author_url": "",
          "post_date": "04/04/2022 06:22:25",
          "content": "<p>Thanks, I don't really know what ArcFace is but I've seen many discussions use this term, I will check it out</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1741006": "Hi, I might be too late to ask this question but I'm new to machine learning and was going through the Happywhale Dataset, I realized that many whales and dolphins have only 1 image per individual in the dataset. I was wondering how are people dealing with this imbalance? \nShould I collect more data?\nShould I just copy the same image of the individual multiple times?\nAny answers would be appreciated 😄",
    "1743653": "The competition context is different that the classical classification or regression problems. Having so many individuals, some with more samples, others with just one  requires a different approach. Using ArcFace the problem is transformed in a resembling recognition more than a clasification. You don't need to worry so much about individual with just one images, focus on image preprocessing, having a good input dataset and post processing techniques on resulting embeddings",
    "1744634": "Thanks, I don't really know what ArcFace is but I've seen many discussions use this term, I will check it out"
  },
  "source": "meta"
}