COLLEGE OF AGRONOMY · HAUDOU GROUP
Rooted in the field. Inspired by discovery.
Understanding crops. Connecting knowledge.
Explore our researchABOUT THE GROUP
HAUWe are Dou Group at the College of Agronomy, Henan Agricultural University. Our research in smart agriculture connects crop and disease patterns in field images, knowledge in agricultural literature, and functional information in plant protein sequences through artificial intelligence.
Meet the group01 RESEARCH
Bringing crop–disease relationships into visual perception to identify diseases and locate lesions in complex field images.
Adapting Qwen3.5-0.8B to agricultural images, with one Chinese category name as the model output.
Connecting image recognition, literature retrieval and optional lesion segmentation for evidence-based agricultural analysis.
Combining protein language models and multi-view features to predict diverse post-translational modification sites in plants.
02 SELECTED PUBLICATIONS
TaxoSeg addresses disease recognition amid similar symptoms and complex field backgrounds by incorporating crop–disease taxonomy into semantic segmentation. Structured semantic prototypes, crop context and symptom descriptions jointly guide disease identification and lesion localization.
Research & resources Molecular Plant 2026PlantPTM predicts diverse post-translational modification sites in plant proteins. Protein language models and multi-view features extract complementary information from sequences to support computational analysis of plant protein function.
Research & resources03 FROM THE GROUP
The group attended the 6th Plant Disease & Pest Remote Sensing Conference in Hangzhou, where Erkun Zhang's TaxoSeg poster received the Best Poster Award.
Predicting diverse plant post-translational modification sites with protein language models and multi-view features. Find the paper, code and prediction platform on our publications page.
Crop–disease taxonomy, structured semantic prototypes and crop context support lesion identification and localization in field images.
Field images were collected at the National Engineering Research Centre for Wheat in Henan. The resulting WheatField-Seg dataset contains 1,762 images with pixel-level lesion annotations for three wheat diseases.