Rice Seed Varieties Hackathon 2025

Challenge Overview

Welcome to the 2025 Rice Seed Varieties Hackathon!

This is an open-ended challenge where teams will develop innovative models and analyses using hyperspectral imaging data of rice seed varieties.

Dataset

We are using the RGB and VIS/NIR Hyperspectral Imaging Data for 90 Rice Seed Varieties dataset:

• Dataset URL: https://zenodo.org/records/3241923

• Description: This comprehensive dataset contains RGB and hyperspectral imaging data for 90 different rice seed varieties, enabling advanced analysis and classification.

• A Small version of the dataset to test on can be found here: https://tinyurl.com/hsi-partial

Important Note: When running any version of the script (Pyhon or Jupyter), the data directory must ALSO be updated in 'helpers.py'. The default path is set up for Maxwell and will need swapping when running locally.

Accessing the Dataset

The dataset has been pre-downloaded onto Maxwell and is available in a shared folder:

/uoa/scratch/shared/2025_hackathon/RGB_and_VIS-NIR_HSI_data_for_90_rice_seed_varieties/RGB_and_VIS-NIR_HSI_data_for_90_rice_seed_varieties

Computing Resources

You will have access to Maxwell's GPU nodes for your model development and analysis.

Example SLURM Script

Attached is an example slurm script "runScript.sh"

Running this will then run the hsipy.py file which will store its ouputs inthe "output" folder.

The job should take a few seconds to run.

Important Note: The --nodelist=agpu004 line will run the job on gpu4. Change the number to use different GPU cards and avoid queueing for the same resource.

Getting Started

1. Copy the example script and modify it for your needs
2. The script will run hsipy.py and store outputs in the "output" folder
3. Expected runtime: A few seconds for the example script

Challenge Goals

Your team's objective is to:

1. Explore the Dataset: Investigate the hyperspectral data to identify interesting patterns
2. Develop a Model: Create an innovative approach to analyze or classify the rice seed varieties
3. Visualize Results: Generate some visualizations that demonstrate your findings
4. Prepare a Presentation: Document your methodology and results for Friday's presentation

Submission Requirements

Your final submission should include:

1. Code Repository: Well-documented code for your analysis and models
2. Results Summary: Key findings and visualizations
3. Presentation: A 15-minute presentation explaining your approach and discoveries
4. Technical Documentation: Methods, challenges, and potential applications

Evaluation Criteria

Teams will be evaluated on:

• Innovation: Originality of approach and techniques
• Technical Merit: Effectiveness and sophistication of models/algorithms
• Insights: Quality and relevance of discoveries from the data
• Presentation: Clarity and engagement of the final presentation

Good Luck!

We look forward to seeing your creative approaches and innovative solutions! If you have any questions, please reach out to the hackathon Christos or Matt.

---