Result

Instructions

The organizers request each team to upload their final or updated trained model, inference code, a ReadMe file, and a detailed description of the algorithm to the competition website by 25th April 2025. Please note: If there is any discrepancy between the submitted results, the provided trained model, and the output generated using the submitted inference code, the team will be disqualified from the competition. In such cases, the team's results will not be displayed on the leaderboard.

Each team is allowed to upload results up to five times per task. The leaderboard will display the best-performing result among all submissions for a task from a team.

Each team member is required to upload the results corresponding to a task.



Submission for Task-A: Handwritten Text Word Detection

Submission of Brief Description of Algorithm
  • Please provide a description (maximum one page) of your proposed method for Task-A. Save the file as one of the following formats: description.doc, description.pdf, description.txt, or description.latex.
Submission of Results
  • The output for Task-A should be saved in a file named as image name.txt (e.g., 328_39691.txt). Each file must include word-level bounding box information in the format: x, y, width, height — with values separated by tabs and arranged in reading order. The output format must exactly match that of the validation set.
  • For each language, create a folder named after the language (e.g., Bengali for the Bengali language), and place all the predicted result files corresponding to the test images of that language inside it. For submissions involving multiple languages, create a parent folder containing ten subfolders — one for each language. Compress the parent folder into a single ZIP file and upload it.
Submission of Trained Model
  • Create a zip file of the trained model and upload
Submission of Inference Code
  • Inference code, requirement.txt with software packages required to run the code, and ReadMe file with instructions for running the code. Make a zip file and upload it.

Select Team Name

 

Upload Result

 

Upload Inference Code

 

Upload Trained Model

 

Upload Brief Description of Algorithm

 

Use any Pretrained Model

 

Use any Additional Training Dataset

 

Use any Synthetic Data for Training

 

Use any Pre-processing Step to Clean the Training Dataset

 

Use any Post OCR Error Correction Technique to Improve OCR Accuracy

 
Submission for Task B: Page Recognition and Reading

Submission of Brief Description of Algorithm
  • Please provide a description (maximum one page) of your proposed method for Task-A. Save the file as one of the following formats: description.doc, description.pdf, description.txt, or description.latex.
Submission of Results
  • There should be one output file corresponding to a page image. The output should be saved as 'image name.txt' (e.g., "120_9135.txt" for page image "120_9135.jpg"), which contains the predicted text of the complete page image in reading order.
  • For each language, create a folder named after the language (e.g., Bengali for the Bengali language), and place all the predicted result files corresponding to the test images of that language inside it. For submissions involving multiple languages, create a parent folder containing ten subfolders — one for each language. Compress the parent folder into a single ZIP file and upload it.
Submission of Trained Model
  • Create a zip file of the trained model and upload
Submission of Inference Code
  • Inference code, requirement.txt with software packages required to run the code, and ReadMe file with instructions for running the code. Make a zip file and upload it.

Select Team Name

 

Upload Result

 

Upload Inference Code

 

Upload Trained Model

 

Upload Brief Description of Algorithm

 

Use any Pretrained Model

 

Use any Additional Training Dataset

 

Use any Synthetic Data for Training

 

Use any Pre-processing Step to Clean the Training Dataset

 

Use any Post OCR Error Correction Technique to Improve OCR Accuracy