The Idea and the Challenge
Pakistan’s agriculture is volatile; the lack of national planning and limited technology results in uneven crop distribution and vulnerability to extreme weather conditions. Spatial planning would potentially improve food security, income, market stability, and resource use—benefiting policymakers, extension services, and farmers.
This research presents a decision-support framework for national crop planning that merges field-level historical, environmental, and socioeconomic data with satellite mosaics and geospatial analysis for crop allocation.
Initial data challenges included:
- Locating field-level historical records and continuous satellite imagery
- Correcting for clouds and sensor issues
- Integrating aerial photos from drones and legacy platforms
- Resolving inconsistent formats, missing data, and scarce ground truth
We developed robust preprocessing pipelines for each sub-problem—harmonization and gap-filling—to create reliable inputs.
The Turning Point
To advance the work, I enlisted Dr. Shoab Ahmed Khan as mentor—bringing cross-domain expertise and leadership crucial to the project’s development. Collaborations through The Center’s network accelerated insights, reduced data-collection time, and enabled rapid prototyping of mini-modules.
A dedicated team at The Center built a mosaicking module that ingested unstable, multi-feed video from a legacy drone system and—via a preprocessing pipeline—stabilized, enhanced, and fused the separated feeds into a super-mosaic for diagnosing issues within agricultural fields.
Collaboration and Environment
The research-driven culture and expert team at The Center made a monumental task far less daunting. Under Dr. Shoab’s supervision—marked by excellence in both research and people—values of integrity and support were instilled across the team. The workplace was enjoyable and safe, particularly for women. As a mother of two young children, I deeply valued the support that allowed me to continue my work with passion and peace of mind.
The Breakthrough — Outcome and Publications
Module 1 — Aerial Preprocessing & Mosaicking
A preprocessing engine for aerial videos with multiple embedded feeds from legacy platforms lacking stabilization.
The system:
- Separated, stabilized, and enhanced each feed
- Generated per-view mosaics
- Combined them into a super-mosaic for field analysis
Module 2 — Yield Prediction (ML)
A machine learning model that ingested historical environmental and non-environmental variables, plus satellite-derived indices, to predict crop yield at specified locations.
Module 3 — Decision-Support Framework
An integrated framework that fused multi-channel historical data—including predicted yields from individual farmers and data from the Ministry of Agriculture and affiliated bodies—to support national crop allocation decisions.
The Lasting Impact — Looking Forward
Personal & Professional Growth. I mastered GIS tools and the agricultural landscape, becoming an out-of-box problem solver ready to operate across practical domains. I strengthened stakeholder engagement, data acquisition, project management, team leadership, scientific writing, and time management—while balancing family responsibilities.
These skills position me for roles in research, policy advisory, and industry focused on data-driven solutions. My time at the Center for Advanced Research in Engineering was foundational: it deepened my passion for applying remote sensing and machine learning to agriculture and equipped me to scale resilient, evidence-based systems nationwide.
Editor
Talha Khan
Tags
Encryption, Data Privacy, Hacking, Big Data (+32)
Sources
-
Nida Rasheed, Waqar S. Qureshi, Shoab A. Khan, Manshoor A. Naqvi, Eisa Alanazi. AirMatch: An Automated Mosaicing System with Video Preprocessing Engine for Multiple Aerial Feeds. IEICE Transactions on Information and Systems (2021) E104.D(4): 490–499. DOI: https://doi.org/10.1587/transinf.2020EDK0003
-
N. Rasheed, S. A. Khan, A. Hassan, S. Safdar. A Decision Support Framework for National Crop Production Planning. IEEE Access (2021) 9: 133402–133415. DOI: https://doi.org/10.1109/ACCESS.2021.3115801