{
  "fetched_at": "2026-10-19T07:38:47Z",
  "source_url": "https://www.kaggle.com/competitions/aii2023-conference-competition/data",
  "pages": [
    {
      "name": "data-description",
      "content": "### **About the data:**\n\n**What files do I need?**  The participant needs images present in the train and test folder, and corresponding train.csv file to train the model. \nNumber of Train samples: 62870\nNumber of Test samples:  28080\n\nFor more details about the dataset, refer:  N. Shaffi and F. Hajamohideen, \"uTHCD: A New Benchmarking for Tamil Handwritten OCR,\" in IEEE Access, vol. 9, pp. 101469-101493, 2021, doi: 10.1109/ACCESS.2021.3096823.\n\n\n**What should I expect the data format to be?**  The image files are in BMP format. The CSV files consists of two columns (ID) and corresponding class label in the second column. \n\n**What am I predicting?**  The class label of the input image consisting of isolated Tamil handwritten character sample \n\n## Files\n\n*   **train.csv** - the training set consists of two columns. The first column (ID) consists of filename and the second column (Class Label) consists of corresponding ground truth label. \n*   **test.csv** - the test set consists of just the filenames in the only column (ID).\n*   **sample_submission.csv** - a sample submission file in the correct format. The Class Label column is mapped to String format (not int value). \n*   **ReadMe.pdf** - supplemental information about the data\n\n## Columns\n*   `ID` - filename of each sample in train/test folder. \n*   `Class Label` - Represents the class label of the Tamil character image. \n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "**About the Problem: Handwritten Tamil Character Recognition**\nWith a rich history spanning over two millennia, the Tamil script holds a significant place in the cul-tural heritage of South India, making the problem of Tamil Handwritten OCR an intriguing challenge. This involves developing technology to accurately recognize and convert handwritten Tamil script in-to digital text, overcoming hurdles posed by the complex nature of the script (Fig.1), variations in handwriting styles (Fig.2) , and limited availability of annotated training data[1]. Despite these chal-lenges, successful development of Tamil Handwritten OCR can unlock a multitude of applications, such as efficient digitization of handwritten documents, automated document analysis and infor-mation extraction, and enhanced accessibility for visually impaired individuals who rely on Tamil script. More details can be found in this research paper [1].\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F9025074%2F89f50577e530ca7c5bf344d6318e24a2%2Ffig11.png?generation=1686235989397566&alt=media)\nFig.1: 156 Character Classes from Tamil Script and their Unicode. \n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F9025074%2Fb68f2cb5070a24c93b17d28652685a79%2Ffig9.png?generation=1686235861622774&alt=media)\nFig.2: Sample Images from the uTHCD database\n\n**References:**\n1) N. Shaffi and F. Hajamohideen, \"uTHCD: A New Benchmarking for Tamil Handwritten OCR,\" in IEEE Access, vol. 9, pp. 101469-101493, 2021, doi: 10.1109/ACCESS.2021.3096823.\n\n2) The uTHCD database homepage, https://www.kaggle.com/faizalhajamohideen/uthcdtamil-handwritten-database\n\n\n\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": "The winning team will have to push their notebooks that reproduce their score. This is mainly because the conference's main theme is: **fostering reproducibility of research results**.\n\nThe winning team will have an option to publish a paper that succinctly presents the solution to the given problem. ",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "About AII2023",
      "content": "This competition is held as part of the 3rd International Conference on Applied Intelligence and Informatics – AII2023 (www.aii2023.org). The conference provides a premier international forum to bring together researchers and practitioners from diverse domains for the sharing of cutting-edge re-search results obtained through the application of intelligence and/or informatics to solve problems that otherwise would not have been possible to solve. AII2023 also fosters the exchange and dis-semination of innovative and practical development of methodologies and technologies with real-life applications. AII2023 asks all authors to make their dataset and methodology open access for com-munity usage. This is to: Foster the Reproducibility of Scientific Results, hence is the theme of the conference. The conference will be held in American University of Ras Al Khaima, UAE known for its stunning natural landscapes, including mountains, deserts, and pristine beaches.\n\n**Important Dates:**\n- Submission deadline: 30th June 2023 (No extension)\n- Winners Announcement: 2nd July 2023  \n- AII2023 in American University of Ras Al Khaima, UAE: 29th  – 31st October 2023\n\n\n**Benefits:** The top three winners of the competition will have the below benefits:\n- Free registration to the conference. \n- A certificate with the indication of the position in the ranking list (1st, 2nd, 3rd)\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}