{
  "fetched_at": "2026-10-19T08:55:10Z",
  "source_url": "https://www.kaggle.com/competitions/garbage-guru-challenge-2-0/data",
  "pages": [
    {
      "name": "data-description",
      "content": "Collection of images of garbages grouped into 10 classes (metal, glass, biological, paper, battery, trash, cardboard, shoes, clothes, and plastic). This data is a subset of a [Garbage dataset](https://www.kaggle.com/datasets/sumn2u/garbage-classification-v2). The number of files in respective classes is as follows: \n\n- **Metal**: 1000\n- **Glass**: 1000\n- **Biological**: 1000\n- **Paper**: 1000\n- **Battery**: 1000\n- **Trash**: 1000\n- **Cardboard**: 1000\n- **Shoes**: 1000\n- **Clothes**: 1000\n- **Plastic**: 1000\n\n\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "**What's New in 2.0?**\n\n**Expanded Dataset:** The dataset now includes more diverse waste categories, improved annotations, and additional real-world images.\n\n**Enhanced Evaluation Metrics:** We've refined our scoring system to better reflect real-world classification performance.\n\n**New Baseline Models:** Participants can access updated starter notebooks to help kickstart their models.\n\n**Improved Labeling Quality:** Data annotations have been enhanced for better accuracy and consistency.\n\n\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Challenge",
      "content": "Craft an AI model that can:\n\n- **Ace the Sorting Exam**: Pinpoint the correct category for each garbage item with impressive precision and recall.\n- **Handle the Messy Real World**: Deal with diverse image variations like different lighting, angles, and cluttered backgrounds.\n- **No More Peek-a-Boo Trash**: Be robust to occlusions and partial visibility of waste items.\n- **Think Beyond the Training Ground**: Generalize well to unseen data and real-world scenarios, becoming a true waste-sorting master.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": "- **Fair Play First**: Cheating is a big no-no! We want to see your genuine skills and creativity in action.  Let's keep it clean and competitive!\n- **Effort Equals Expertise**: Apply yourself! This is a learning journey, so dive deep, experiment, and refine your models. Don't just throw spaghetti at the code wall. The more effort you put in, the more you'll get out of this challenge. Remember, practice makes perfect (or at least a highly accurate AI model)!\n- **Fun at the Forefront**: While we're tackling a serious issue, let's not forget to have fun! Enjoy the process of learning, share your insights with the community, and celebrate each other's accomplishments. Turning trash into treasure should be a blast!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "abstract",
      "content": "Welcome to the second edition of the **Garbage Guru Challenge**! Building on the success of the first competition, we aim to push the boundaries of AI-powered waste classification and recycling technology. This challenge provides an opportunity for participants to leverage machine learning techniques to create robust models for identifying and sorting different types of waste efficiently.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Dataset",
      "content": "- The dataset consists of images labeled with different waste categories such as plastic, paper, glass, metal, organic, and more.\n- Each image is accompanied by its corresponding class label.\n- A detailed data description and sample code are provided to help you get started.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Rewards",
      "content": "- **Knowledge is Power**: Gain valuable experience and insights into AI for environmental applications.\n- **Sharpen Your Portfolio**: Showcase your skills and expertise to potential employers and collaborators.\n- **Contribute to a Greener Future**: Your learnings can pave the way for smarter waste sorting systems, promoting a more sustainable world.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "- **Macro F1 score**: The heart of the matter! We'll measure how often your model correctly classifies each type of garbage using Macro F1-Score. See https://scikit-learn.org/stable/modules/generated/sklearn.metrics.f1_score.html.\n- **Robustness**: Can your model handle real-world challenges like blurry images and mixed waste streams? We'll test its adaptability.\n- **Interpretability**: Bonus points for models that explain their reasoning, helping us understand and improve their decision-making.\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Bonus Quests ",
      "content": "- **Transfer Learning Mastery**: Utilize pre-trained image classification models and fine-tune them for garbage sorting.\n- **Data Augmentation Guru**: Unleash the power of data augmentation techniques to boost your model'sgeneralizability.\n- **Explainable AI**: Develop models that reveal their reasoning behind each classification, fostering trust and understanding.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Submission",
      "content": "For each id in the test set, you must predict a type of class (or label). The file should contain a header and have the following format:\n\n```csv\nId,Expected\npaper/paper_1775.jpg,0\npaper/paper_1749.jpg,0\n.....\n```\n\nThe expected numeric class labels are assigned based on a predefined order rather than alphabetical sorting. The class labels are mapped as follows:\nPaper → 0\nClothes → 1\nMetal → 2\nCardboard → 3\nTrash → 4\nGlass → 5\nBiological → 6\nBattery → 7\nPlastic → 8\nShoes → 9",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Summary",
      "content": "Are you ready to take on the challenge and make an impact on sustainability? Join now and be part of the movement towards smarter waste management!\n\nLet’s innovate for a cleaner planet! 🌍♻️",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}