Skip to content

ProblemStatement.md

Latest commit

 

History

History
64 lines (45 loc) · 3.15 KB

ProblemStatement.md

File metadata and controls

64 lines (45 loc) · 3.15 KB
Raw

Multi-QR Code Recognition for Medicine Packs

Problem Statement

This hackathon is about solving a real problem in the medical field – recognizing multiple QR codes on medicine packs.
Most medicine packs have more than one QR code, such as codes for the manufacturer, batch number, distributor, or regulator. The challenge is to detect all these QR codes from an image. For advanced participants, an additional bonus challenge is to decode the QR contents and classify them.

Dataset

Dataset

  • You will be given a dataset of 200 medicine pack images with annotated QR bounding boxes.
  • These will help you train and validate your models for the main detection task.
  • A test set of 50 images will be used for final scoring, which also contains decoded QR values for bonus evaluation.
  • Additional hidden images will also be used to validate accuracy and benchmarking.

What Needs to be Done

  • The submitted code must be complete, runnable, and reproducible.
  • Detect all QR codes in the provided images using the bounding box annotations for training.
  • Ensure detection works even with tilted, blurred, or partially covered images.
  • For those achieving strong detection performance, attempt the bonus task of decoding and classifying the QR codes on the test set.

What Makes it Competitive

  • Main challenge: QR detection – all participants compete on the same dataset.
  • Bonus challenge: QR decoding and classification – available only after demonstrating strong detection accuracy.
  • Dataset includes realistic variations such as lighting, tilt, and occlusion.
  • Extra credit for efficient solutions (lighter, faster models rank higher in case of ties).

Summary

  • Main Task → Detect all QR codes in medicine pack images.
  • Bonus Task → Decode and classify QR codes.
  • Evaluation → Based on detection accuracy, robustness, efficiency, and bonus classification performance.

Evaluation Plan

Restrictions

  • Do not use any external APIs (including QR code reading/decoding APIs) at any stage for generating results.
  • The submitted code must be complete, runnable, and reproducible.

Organizers should be able to:

  1. Clone/unzip your repository.
  2. Follow the README instructions.
  3. Run a single command to reproduce your output on the given images.

Scoring

1. Decoding & Classification Score (Bonus Task)

  • Metric: String accuracy of decoded QR values + correct classification type.
  • Weight: Added bonus on hidden 50 images only.

2. Final Score Composition

  • Participants train and test on the provided 200 images.

  • They must submit:

    • submission_detection_1.json → JSON file for the test 50 images.
    • Complete runnable code + README.

  • Organizers will then run the submitted code on 50 hidden images (not shared with participants) to evaluate:

    • Detection (mandatory).
    • Decoding & Classification (bonus).

  • Winners will be decided based on hidden test performance.