Logo Lanfrica
  • Home
  • Atlas
  • Insights
  • Docs
  • Sign in

© 2026 Lanfrica. All rights reserved. All copyrights of the resources shown on the Lanfrica website belong to the original copyright holders, unless explicitly stated otherwise.

Collins-67/uav-crop-mapping-mozambique

Domain:

agriculturegeospatial

Record type:

paper
Creator:
Col
Host:
SegFormer-based crop type mapping and FCover estimation from UAV imagery in smallholder landscapes of Mozambique. MSc Thesis — ITC, University of Twente, 2026. # UAV-based Crop Mapping and FCover Estimation ### Smallholder-Dominated Landscapes of Mozambique **Author:** Collins Edem Hlordzie **Institution:** ITC, Faculty of GEO, University of Twente **Supervisors:** Dr. F. J. Ellsäßer | Dr. F. B. Osei **Year:** 2026 --- ## Overview This repository contains the analysis code for my MSc thesis evaluating UAV-based RGB and multispectral imagery with SegFormer deep learning for crop type mapping and fractional vegetation cover (FCover) estimation across 19 sampling areas in Mozambique. ## Research Questions 1. How accurately can crop areas be segmented from non-crop areas and compared with enumerator data? 2. How do UAV-derived FCover estimates differ from enumerator-based assessments for the identified crop types? ## Key Findings - MS model significantly outperforms RGB (McNemar p<0.001, OR=1.483) - MS Maize FCover is the only class showing significant positive agreement with enumerator data (r=0.178**, ρ=0.239***) ## Notebook Execution Order | Step | Notebook | Description | |------|----------|-------------| | 1 | 01_label_generation.ipynb | Generate label rasters | | 2 | 02a_RGB_tiling.ipynb | Tile RGB imagery | | 3 | 02b_MS_tiling.ipynb | Tile MS imagery | | 4 | 03a_data_split.ipynb | Train/val/test split | | 5 | 04_RGB_training.ipynb | Train RGB model | | 6 | 04b_MS_training.ipynb | Train MS model | | 7 | 05_inference_and_fcover.ipynb | Inference + FCover | | 8 | 06_full_19_inference.ipynb | Full 19-SA inference | | 9 | 07a_agricultural_statistics.ipynb | Agri statistics | | 10 | 07b_agri_figures.ipynb | Figures | | 11 | 08_inferential_statistics.ipynb | Statistical tests | ## Data Access Raw UAV orthomosaics and enumerator field data are restricted under the ITC research programme data-sharing agreement. Contact: rdm-itc@utwente.nl ## Requirements pip install -r requirements.txt ## Citation Hlordzie, C. E. (2026). UAV-based crop mapping and fractional vegetation cover estimation in smallholder-dominated landscape …

Visit

github.com

Tasks

image classificationcomputer vision

Tags

cropcropmappingdeeplearningmozambiqueprecision-agriculturepythonpytorchremote-sensingsegformersegmentation+1

Licenses

MIT