Applied research and scientific methods
Experimental design, quantitative analysis, critical interpretation and methodological evaluation.
Bruna Condé Lima, PhD is a researcher and consultant. Her work connects experimental science and computer vision with remote sensing, geospatial analysis and machine learning. Recent applications are agricultural, but the approach is not limited to one crop or sector.
Interior excerpts from a real monitoring project; no field boundaries or geographic references.
Her experience combines scientific study design and computational methods to answer specific questions. Agricultural case studies provide practical evidence of approaches that can be applied in other settings.
Experimental design, quantitative analysis, critical interpretation and methodological evaluation.
Published experience in image analysis and recent semantic-segmentation research using deep learning.
Processing and interpretation of UAV and Sentinel imagery to examine spectral responses and changes over time.
Relative zoning, time series and anomaly analysis, with reproducible maps and GIS layers.
Recent case studies and research covering rice and vineyards. The underlying methods may be explored across other crops, regions and sectors.
Case study: rice, 2026 season. UAV imagery and Sentinel observations combined to interpret relative vegetation response, changes over time and within-field variability.
Spectral indices, three-class relative zoning and change analysis across flights.
Sentinel-1/Sentinel-2 monitoring and relative anomaly assessment.
Maps, GIS layers and interpretative reports to guide field inspection.
Real anonymised excerpts; client boundaries, coordinates and complete reports are not published.
Sentinel time series and vegetation/moisture indices used to study how land varies over time. Recent application: vineyards and parcel assessments supporting land-related decisions.
Experimental deep-learning semantic-segmentation work supported by high-performance computing resources (Leonardo HPC), with initial applications to UAV imagery in rice. Collaboration is welcomed from agricultural companies, UAV operators and research groups interested in exploring the approach across different regions, crops and field conditions.
Propose a collaboration ↗The initial research question or decision determines which data sources, analytical methods and checks are suitable, rather than starting with a fixed list of indices.
The underlying question, decision or research objective is clarified, together with the available evidence.
Imagery, time series and reference observations are checked for quality and suitability.
Appropriate methods are selected and critically evaluated in light of their limitations and the context.
Findings are presented in an accessible form, with maps, GIS layers, code or reports where appropriate.
Every project communicates data limitations and the level of confidence supported by the evidence.
Bruna Condé Lima, PhD, is a researcher and consultant trained in food engineering, with a doctorate from the University of Melbourne. Her background connects experimental research, analytical methods and computer vision.
Her academic work included sparkling-wine quality, analytical chemistry, and assessment using robotic systems and image analysis. Related research appeared in journals including Food Control and Food Research International.
This foundation evolved into deep learning and out-of-distribution image recognition with VitiNet (2025), followed by multispectral imagery, Sentinel-1/2 observations and geospatial analysis. Agricultural projects demonstrate capabilities that can also be applied across other crops and sectors.
PhD in Agricultural ScienceUniversity of Melbourne
MSc in Food ScienceUniversity of Melbourne
BSc in Food EngineeringUniversidade Federal de Viçosa
Research and university teachingUniversity of Melbourne; peer-reviewed publications
Applied ML trainingBSC AI Factory · Essential Machine Learning (completed)
Peer-reviewed work in food science, image analysis, robotics and analytical methods.
VitiNet: ResNet18-based grape-leaf disease classification with out-of-distribution (OOD) image recognition.
UAV imagery, Sentinel observations, time series and deliverables for agricultural monitoring.
Semantic-segmentation research for weed mapping, supported by Leonardo HPC resources.
Selected journal papers and a conference contribution tracing a progression from experimental research and computer vision to applied deep learning.
Food Research International
View publicationFood Control
View publicationJournal of Agricultural and Food Chemistry
View publicationCritical Reviews in Food Science and Nutrition
View publicationIECAG 2025 · 5th International Electronic Conference on Agronomy
ResNet18-based grape-leaf disease classifier designed to recognise out-of-distribution (OOD) images. On an experimental test set it achieved 98.2% overall accuracy and 99.6% precision for the “Other” class. These results describe the study setting, not field-validated performance.
View conference contributionFor enquiries about applied research, technical consulting and methodology development in remote sensing, geospatial analysis, computer vision and machine learning.