CGMap: A Geospatially Aware Deep Learning Framework for Crop Gap Mapping Using UAV

Manual inspection of crop germination is labor-intensive and error-prone, yet critical for identifying "bald spots"—areas where missing plants significantly reduce potential yield. This study addresses the need for a scalable, automated, and geospatial-aware method to monitor early-stage sugarcane germination using drone-based imagery.

The authors propose a pipeline that integrates object detection with spatial analysis. First, they use the YOLOv8 architecture to detect individual sugarcane seedlings from high-resolution UAV orthomosaics. To handle variations in planting geometry, the researchers introduced a novel orientation-normalization technique using Minimum Spanning Trees (MST). By constructing a graph of detected plant centroids and filtering connections based on expected inter-row spacing, the system can reliably extract row and column structures even in off-axis or irregular fields. Finally, the system identifies gaps by comparing the actual spacing between detected plants against expected planting intervals, exporting the results as georeferenced maps in Well-Known Text (WKT) format.

The pipeline successfully automates the identification of germination gaps, generating actionable maps within 30 minutes of processing. By providing precise locations of missing plants, the system enables targeted transplantation, which can help mitigate the 10–15 tonnes per hectare yield loss often associated with poor germination. The modular design allows the framework to be integrated into existing GIS platforms, making it a practical tool for sugar mills and agronomists to manage field productivity and resource allocation more efficiently.

This research provides a scalable solution for precision agriculture in large-scale sugarcane production. By moving from manual, subjective inspections to data-driven, automated monitoring, farmers can optimize their crop stands early in the growth cycle. The methodology is not strictly limited to sugarcane and shows potential for adaptation to other row crops with defined planting patterns, contributing to more sustainable and profitable agricultural practices.