Successful COLMAP commands for low-altitude drone collect

[rickym...@gmail.com]

I just wanted to share these so others can have a quick start to testing & making use of COLMAP -- especiallly in the case of low-altitude UAS.

18 images were used from a Sensefly eBee with a Canon Powershot S110, 4048 x 3048, 5mm focal length. Images were placed in $DATASET_PATH/img

Processing was done on a Dell Precision 7510, Intel Core i7-6820HQ CPU_2.7GHz with an Nvidia Quadro M2000M GPU running Ubuntu 18.04:

DATASET_PATH=/path/to/data/folder/

colmap feature_extractor --database_path $DATASET_PATH/database.db --image_path $DATASET_PATH/img --ImageReader.camera_model "RADIAL" --SiftExtraction.max_image_size 5000

colmap exhaustive_matcher --database_path $DATASET_PATH/database.db

mkdir $DATASET_PATH/sparse

colmap mapper --database_path $DATASET_PATH/database.db --image_path $DATASET_PATH/img --output_path $DATASET_PATH/sparse --Mapper.max_focal_length_ratio 10 --Mapper.max_extra_param 1000000

mkdir $DATASET_PATH/dense

colmap image_undistorter --image_path $DATASET_PATH/img --input_path $DATASET_PATH/sparse/0 --output_path $DATASET_PATH/dense --output_type COLMAP --max_image_size 5000

colmap patch_match_stereo --workspace_path $DATASET_PATH/dense --workspace_format COLMAP --PatchMatchStereo.max_image_size 2000

colmap stereo_fusion --workspace_path $DATASET_PATH/dense --workspace_format COLMAP --input_type geometric --output_path $DATASET_PATH/dense/fused.ply

colmap poisson_mesher --input_path $DATASET_PATH/dense/fused.ply --output_path $DATASET_PATH/dense/meshed-poisson.ply

colmap delaunay_mesher --input_path $DATASET_PATH/dense --output_path $DATASET_PATH/dense/meshed-delaunay.ply