Research profile ❧ AI × Plant Pathology
Колесников Никита Сергеевич
PhD student at Volgograd State Technical University. I work on AI methods for plant disease diagnosis that survive the journey from the lab to the field — and fit on a device you can hold in one hand.
CORAL, CW-CORAL, DANN and CDAN for plant disease recognition when training images come from the lab and real leaves come from the field.
Stage-wise attribution of accuracy loss in compression pipelines; INT8 post-training quantization; ~10× feature database compression without retrieval loss.
Running diagnosis on Raspberry Pi-class hardware: latency budgets, hardware benchmarking infrastructure, deployable models under 3 MB.
Grad-CAM and SHAP explanations for diagnosis models, aimed at decisions an agronomist can actually check.
A System Model of Edge Plant Disease Diagnosis: 10× Feature Database Compression Without Loss of Retrieval Accuracy
Системная модель edge-диагностики заболеваний растений: 10-кратное сжатие базы признаков без потери точности поиска
SEDM-2026 conference proceedings · accepted, in press, 2026