Systematic Review · 2026
Trends and Innovations in Modern Agriculture
1 / 10
Babenko Y. · Academia Engineering 2026

Trends and Innovations
in Modern Agriculture

From fluorescent sensors to decision support systems — a systematic review of 65 studies mapping the Plant-IT architecture.

65 studies reviewed 5 query streams 5 Plant-IT layers

PRISMA Screening Workflow

19,217
Records Identified
5 queries (Q1–Q5) in Scopus
15,163
After Scopus Filters
AGRI, ENGI, COMP · Articles, Reviews · English
70
Eligible (before dedup)
Domain-expert full-text screening
65
Final Corpus
5 duplicates removed · mapped to Plant-IT layers

Plant-IT 5-Layer Architecture

5
DSS Layer
Decision support & prescriptions · 8 studies (12.3%)
4
IoT Integration Layer
Data transmission & edge-cloud · 6 studies (9.2%)
3
Machine Learning Layer
Pattern recognition & prediction · 12 studies (18.5%)
2
Computer Vision Layer
Image-based phenotyping & disease detection · 17 studies (26.2%)
1
Sensor & Plant Layer
In situ & remote sensing · 22 studies (33.8%)
Click a layer to jump to its details →

Sensor & Plant Layer (22 studies)

Hyperspectral Imaging
Drought stress, nutrient status, disease detection
Thermal Cameras
Water stress, transpiration, irrigation optimization
CFI / Fluorimeters
Photosynthetic efficiency, early stress (Kautsky effect)
LiDAR
3D canopy structure, biomass estimation
VOC Gas Sensors
Volatile organic compounds for pathogen detection
Wearable Sensors
Continuous leaf/stem monitoring, micro-environment
Electrochemical Biosensors
Metabolites, phytohormones, heavy metals
Acoustic / Ultrasound
Growth dynamics, internal tissue damage

Computer Vision Layer (17 studies)

CNN
PlantVillage CNN
38 disease classes · leaf-level classification
YOLO
YOLOv5/v8/v11
Real-time multi-class plant health monitoring · spot-level detection
ViT
Vision Transformers
Complex/noisy backgrounds · global contextual info
GAN
GAN Augmentation
Synthetic data for rare diseases · improves generalization
UAV
UAV + Multi-modal Fusion
RGB + thermal + hyperspectral · large-scale spatial coverage

Key Vegetation Indices

IndexMeasuresApplication
NDVIChlorophyll, biomassCrop vigor, drought, yield prediction
GNDVIChlorophyll concentrationNitrogen deficiency, early stress
EVILAI, canopy structureDense vegetation, atmospheric correction
NDREPhotosynthetic efficiencyNitrogen status, late-season monitoring
PRILight use efficiencyShort-term stress, photosynthetic dynamics
SAVIVegetation coverSparse vegetation, arid zones

IoT Integration & Decision Support

IoT Architecture
Sensor nodes (CFI, spectral, thermal, soil)
LoRaWAN / Zigbee / NB-IoT
Edge + cloud computing
Dashboards & mobile apps
Blockchain for data integrity
Digital Twins
Virtual plant models
Real-time sensor updates
Predict crop responses
Test interventions virtually
Scalable: plot → commercial farm
Bio-Inspired DSS
Swarm intelligence (ant colony)
Neuro-symbolic hybrids
Fuzzy logic (bee foraging)
Plant signal propagation
Adaptive, noise-tolerant
Key Bottlenecks
Data heterogeneity & standards
Device interoperability
Energy constraints
Cybersecurity risks
Limited closed-loop control

Temporal Evolution of Plant-IT

1990s
Sensor & physiology research emerges (Q1)
2002+
DSS-oriented research begins (Q3)
2021–22
Remote sensing inflection — UAVs & satellites (Q4)
2022–23
Sensor research accelerates — 133 pubs in 2023 (Q1)
2025
Precision agriculture explodes — 1566 pubs (Q2)
Steady
Computer vision — gradual maturation, 10–20 pubs/yr (Q5)
Bottom-up evolution: sensing → perception → learning → decision

Key Findings & Gaps

Fragmentation
Most studies address isolated components — rarely integrated end-to-end
Standardization Gap
No unified data formats, ontologies, or metadata — limits transferability
Field Robustness
CV models struggle with illumination, background, domain shift
Closed-Loop Rarity
AI mostly for classification — automatic irrigation/nutrient dosing uncommon
Knowledge Divide
Agronomists lack DSS literacy; engineers lack plant physiology — bridge needed

Conclusions & Future Directions

1
Plant-IT architecture provides a conceptual framework for integrating sensing → analytics → decision support
2
Harmonized standards are critical — semantic frameworks, compatible protocols, FAIR principles
3
AI-on-the-edge — energy-efficient preprocessing closer to sensors for real-time responsiveness
4
Digital twins + bio-solutions — coupling microbial consortia, biopesticides with real-time monitoring
5
Adaptive HMIs — human-centric DSS interfaces that adjust to context, cognitive load, and crop stage
Babenko Y. · Academia Engineering 2026 · DOI: 10.20935/AcadEng8153
100%