Congratulations to Lin Yu-Fan, Zheng Jing-Heng, Qiu Bo-Cheng, Kang Cheng-Jun, and Li Chia-Ming for winning first place in the ICPR Automated Crop Disease Diagnosis from Hyperspectral Imagery 3rd Challenge.

This competition required participants to use a large dataset of multispectral/hyperspectral remote sensing images to improve the accuracy and efficiency of crop disease identification, contributing to sustainable agricultural practices and global food security. The challenge was particularly demanding due to the very small spatial resolution of the images (only 20x20 pixels), requiring differentiation of crop conditions over large areas while avoiding overfitting.
The team designed a semi-supervised learning framework based on hyperspectral characteristics, utilizing both Endmembers and key features for predictions. They stood out among teams from various countries and achieved global champion status.
Multispectral and hyperspectral images indeed have unique properties compared to regular images, and understanding them took considerable time. However, the hard work over the past year or two paid off, allowing the team to quickly master the challenge. Congratulations once again!

Yu-Fan Lin, Ching-Heng Cheng, Bo-Cheng Qiu, Cheng-Jun Kang, Chia-Ming Lee, Chih-Chung Hsu, "Self-supervised Fusarium Head Blight Detection with Hyperspectral Image and Feature Mining", ICPR 2024 Competition: Beyond Visible Spectrum: AI for Agriculture
CODE and Results will be released soon.
Official Website: https://han-research.gitlab.io/Agvision/