{
  "id": 583488,
  "title": "A Kaggle-Friendly MASt3R Demo Notebook for Robust Image Matching!",
  "url": "/competitions/image-matching-challenge-2025/discussion/583488",
  "author_name": "Helen",
  "post_date": "2025-06-07T08:23:24.996000",
  "votes": 2,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi everyone,</p>\n<p>I'm excited to share a new notebook I've been working on: <a href=\"https://www.kaggle.com/code/yuanlin08/mast3r-gradio-demo\" target=\"_blank\">Mast3R-Gradio demo</a></p>\n<p>Inspired by the excellent work in the <a href=\"https://www.kaggle.com/competitions/image-matching-challenge-2025/discussion/583058\" target=\"_blank\">first-place solution</a> by <a href=\"https://www.kaggle.com/ns6464\" target=\"_blank\">@ns6464</a> , I've adapted the MASt3R model for a more streamlined and accessible experience directly within Kaggle notebooks.</p>\n<p>This demo focuses on showcasing MASt3R's powerful image matching capabilities. In the notebook, you'll find:</p>\n<ol>\n<li><p>Clear setup for Kaggle environments (offline mode) for MASt3R.</p></li>\n<li><p>Gradio integration (with public sharing enabled): The notebook includes a simple Gradio interface so you can interact with the model directly via a public link (when --share is enabled!, and make sure to turn on the internet to do this).</p></li>\n</ol>\n<p>My goal was to provide a straightforward example for anyone interested in diving into MASt3R's features without battling complex local setups.</p>\n<p>Feel free to fork the notebook, experiment with your own images, and share your thoughts!</p>",
  "messages": [
    {
      "id": 3219162,
      "postDate": "2025-06-07T08:23:24.997Z",
      "content": "<p>Hi everyone,</p>\n<p>I'm excited to share a new notebook I've been working on: <a href=\"https://www.kaggle.com/code/yuanlin08/mast3r-gradio-demo\" target=\"_blank\">Mast3R-Gradio demo</a></p>\n<p>Inspired by the excellent work in the <a href=\"https://www.kaggle.com/competitions/image-matching-challenge-2025/discussion/583058\" target=\"_blank\">first-place solution</a> by <a href=\"https://www.kaggle.com/ns6464\" target=\"_blank\">@ns6464</a> , I've adapted the MASt3R model for a more streamlined and accessible experience directly within Kaggle notebooks.</p>\n<p>This demo focuses on showcasing MASt3R's powerful image matching capabilities. In the notebook, you'll find:</p>\n<ol>\n<li><p>Clear setup for Kaggle environments (offline mode) for MASt3R.</p></li>\n<li><p>Gradio integration (with public sharing enabled): The notebook includes a simple Gradio interface so you can interact with the model directly via a public link (when --share is enabled!, and make sure to turn on the internet to do this).</p></li>\n</ol>\n<p>My goal was to provide a straightforward example for anyone interested in diving into MASt3R's features without battling complex local setups.</p>\n<p>Feel free to fork the notebook, experiment with your own images, and share your thoughts!</p>",
      "rawMarkdown": "Hi everyone,\n\nI'm excited to share a new notebook I've been working on: [Mast3R-Gradio demo](https://www.kaggle.com/code/yuanlin08/mast3r-gradio-demo)\n\nInspired by the excellent work in the [first-place solution](https://www.kaggle.com/competitions/image-matching-challenge-2025/discussion/583058) by @ns6464 , I've adapted the MASt3R model for a more streamlined and accessible experience directly within Kaggle notebooks.\n\nThis demo focuses on showcasing MASt3R's powerful image matching capabilities. In the notebook, you'll find:\n\n1. Clear setup for Kaggle environments (offline mode) for MASt3R.\n\n2. Gradio integration (with public sharing enabled): The notebook includes a simple Gradio interface so you can interact with the model directly via a public link (when --share is enabled!, and make sure to turn on the internet to do this).\n\nMy goal was to provide a straightforward example for anyone interested in diving into MASt3R's features without battling complex local setups.\n\nFeel free to fork the notebook, experiment with your own images, and share your thoughts!",
      "votes": 2
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "3219162": "Hi everyone,\n\nI'm excited to share a new notebook I've been working on: [Mast3R-Gradio demo](https://www.kaggle.com/code/yuanlin08/mast3r-gradio-demo)\n\nInspired by the excellent work in the [first-place solution](https://www.kaggle.com/competitions/image-matching-challenge-2025/discussion/583058) by @ns6464 , I've adapted the MASt3R model for a more streamlined and accessible experience directly within Kaggle notebooks.\n\nThis demo focuses on showcasing MASt3R's powerful image matching capabilities. In the notebook, you'll find:\n\n1. Clear setup for Kaggle environments (offline mode) for MASt3R.\n\n2. Gradio integration (with public sharing enabled): The notebook includes a simple Gradio interface so you can interact with the model directly via a public link (when --share is enabled!, and make sure to turn on the internet to do this).\n\nMy goal was to provide a straightforward example for anyone interested in diving into MASt3R's features without battling complex local setups.\n\nFeel free to fork the notebook, experiment with your own images, and share your thoughts!"
  }
}