%0 Report
%A Yildiz, Erenus
%T 3D Reconstruction of Cassava Roots Using COLMAP
%M FZJ-2025-04413
%P 46p
%D 2025
%X Cassava (Manihot esculenta) is an important crop for food security in tropical and subtropical regions, with roots containing up to 85% starch on a dry weight basis [25]. Understanding root system architecture is essential for breeding programs aimed at improving yield and stress tolerance [1]. Traditional 2D imaging methods for root phenotyping have limitations in capturing threedimensional root structures, leading to incomplete trait measurements [28, 8]. While advanced 3D methods like DIRT/3D 2.0 exist [14], they require specialized equipment that may not be accessible to all research facilities. This laboratory work implemented a cost-effective 3D reconstruction pipeline using standard DSLR cameras and open-source software (REMBG, COLMAP) to extract morphological traits from cassava roots. The pipeline combined deep learning-based segmentation with structure from motion techniques to analyze 1039 cassava root system acquisitions from plants aged 5-10 weeks collected at the Rayong Field Crops Research Center in Thailand, achieving 62% reconstruction success (644 successfully reconstructed roots).
%F PUB:(DE-HGF)15
%9 Internal Report
%R 10.34734/FZJ-2025-04413
%U https://juser.fz-juelich.de/record/1047611