Sohail Anwar
I build machine learning systems that solve real agricultural and environmental problems: predicting nutrient content in silage with near‑infrared spectroscopy, detecting vineyard abandonment from satellite imagery, and reconstructing 3D tree structure from low‑cost stereo cameras to support automated pruning.
From Jamshoro to Pavia, building AI for problems that matter
I am a PhD researcher in the Department of Electrical, Computer and Biomedical Engineering at the University of Pavia (expected March 2027), where my work sits at the intersection of machine learning, remote sensing, and sustainable agriculture. My research is part of NODES (Nord Ovest Digitale e Sostenibile), an initiative funded by Italy’s Ministry of University and Research (MUR) under the PNRR to drive digital transformation across North‑West Italy.
I grew up in a working‑class household in Jamshoro, Pakistan, and was the first in my family to pursue an engineering university education. I completed my Bachelor’s and Master’s in Electronic Engineering at Mehran University of Engineering & Technology, where I built a deep learning system for cotton crop disease detection that reached 98.6% accuracy on a Raspberry Pi prototype. Before my PhD, I worked as a Research Assistant at Mehran University of Engineering and Technology, contributing to a SHEC-funded project on IoT-based cotton crop disease detection, and led IoT courses under Pakistan’s NAVTCC program. I then moved to Italy to pursue doctoral research in agricultural AI.
As part of the mandatory abroad period of my PhD, I am currently a visiting researcher at Universidad Loyola Andalucía in Seville, developing 3D structural reconstruction of olive trees for automated pruning. I also serve as a teaching assistant for Industrial Control at the University of Pavia and for Dynamical Systems for Industrial Automation at the University of Milan. I am currently learning Italian (targeting B1) and Spanish, and speak English, Sindhi and Urdu fluently.
Where I work
Three threads run through my research: agricultural sensing, geospatial computer vision, and 3D structural modelling for robotics.
Agricultural Machine Learning
Developing and validating ML models for the FORMIDABILÆ project and for full‑scale biogas plants, predicting nutrient content and biogas quality from near‑infrared spectroscopy, process data, and microbial community composition.
NIR Spectroscopy PLS Regression SHAP
Remote Sensing & Computer Vision
Leading the VINO project’s use of satellite imagery and YOLOv8 to detect vineyard condition and abandonment across the Oltrepò Pavese wine region, generating suitability maps that support climate‑resilient viticulture.
YOLOv8 Google Earth Engine QGIS
3D Reconstruction & Structural Modelling
As a visiting researcher at Universidad Loyola, building a pipeline from a low‑cost stereo depth camera to a branch‑level structural model of a tree, to support automated pruning decisions on olive, a species prior robotic‑pruning work has avoided.
Stereo Vision Point Clouds Finite‑Element Modelling
Featured Projects
Full write‑ups of the technical approach, challenges, and results for each project.
Cotton Crop Disease Detection
An end‑to‑end deep learning system identifying Bacterial Blight and Cotton Leaf Curl Virus from leaf images, deployed on a Raspberry Pi for real‑time field use.
PyTorch Raspberry Pi Edge AI
Vineyard Condition Detection
A YOLOv8 pipeline that detects active vs. abandoned vineyards from Google Earth imagery across the Oltrepò Pavese wine region, presented at EGU General Assembly 2025.
YOLOv8 Remote Sensing Geospatial ML
NIR Spectroscopy ML Pipeline
A regression pipeline turning raw near‑infrared spectra into calibrated nutrient predictions for maize, dairy, and biogas systems, replacing days‑long lab tests with seconds‑long scans.
Scikit‑learn XGBoost PLS Regression
Current Research
Two further lines of work, underway now and heading toward publication.
3D Structural Modelling of Olive Trees
Reconstructing branch‑level structural models of olive trees from a low‑cost stereo depth camera, to support automated pruning on a species that is always pruned in full leaf.
ML for Biogas Plant Optimization
Predicting energy output, biogas quality, and reactor instability in a full‑scale anaerobic digestion plant from process data and microbial community composition.
Recent Research
JOURNAL · 2026
An exploratory feasibility machine learning study to assess fermentative quality in silage
Novara, V., Marchese, M., Anwar, S., Toffanin, C., & Gallo, A., Animal Feed Science and Technology, article 116990.
JOURNAL · 2025
Soil Management and Machine Learning Abandonment Detection in Mediterranean Olive Groves Under Drought
Marchese, G., Herranz‑Luque, J. E., Anwar, S., Vaglia, V., Toffanin, C., et al., Soil Systems, 9(4), 118.
Conferences & Speaking
Presenting research at international venues in 2025, including EGU General Assembly in Vienna and TERRAENVISION in Granada.
Let’s work together
Open to research collaborations, visiting positions, and applied ML consulting.
- ☎ +39 393 598 1427
- ✉ sohail.anwar01@universitadipavia.it
- 📍 Pavia, Italy