ERROR: [main_ComputeMatches.cpp:289] Invalid Nearest Neighbor method: ANNL2

[Ocezo]

Hello Pierre Moulon (@pmoulon) and all the contributors,

First of all, thanks a lot for the huge work on openMVG! I would really make it work! Unfortunately it's not the case yet...

I'm on Ubuntu 24.04 LTS. I followed the building instructions for Linux. I put the CMake option "-DOpenMVG_BUILD_TESTS=ON" and ran the tests with "make test". I got the final green sentence : "100% tests passed, 0 tests failed out of 84".

Then I moved to "openMVG_Build/software/SfM" and run the demo tutorial in Python. I got the error "Invalid Nearest Neighbor method: ANNL2"... Where can I configure the NN method ? Are there other things to tune ? Can you please help me ?

To let you know, here's my terminal:

jm@hal:~/Documents/openMVG_Build/software/SfM$ python tutorial_demo.py 
Clonage dans 'ImageDataset_SceauxCastle'...
remote: Enumerating objects: 17, done.
remote: Total 17 (delta 0), reused 0 (delta 0), pack-reused 17 (from 1)
Réception d'objets: 100% (17/17), 12.05 Mio | 2.91 Mio/s, fait.
Using input dir  :  /home/jm/Documents/openMVG_Build/software/SfM/ImageDataset_SceauxCastle/images
      output_dir :  /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out
1. Intrinsics analysis
INFO: [main_SfMInit_ImageListing.cpp:194]  You called : /home/jm/Documents/openMVG_Build/Linux-x86_64-RELEASE/openMVG_main_SfMInit_ImageListing
--imageDirectory /home/jm/Documents/openMVG_Build/software/SfM/ImageDataset_SceauxCastle/images
--sensorWidthDatabase /home/jm/Documents/openMVG/src/software/SfM/../../openMVG/exif/sensor_width_database/sensor_width_camera_database.txt
--outputDirectory /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches
--focal -1
--intrinsics 
--camera_model 3
--group_camera_model 1
--use_pose_prior 0
--prior_weights 1.0;1.0;1.0
--gps_to_xyz_method 0
INFO: [loggerprogress.hpp:79] [- Listing images -] 100%
WARNING: [main_SfMInit_ImageListing.cpp:455] Warning & Error messages:
K.txt: Unkown image file format.
Readme.txt: Unkown image file format.

INFO: [main_SfMInit_ImageListing.cpp:475] SfMInit_ImageListing report:
listed #File(s): 13
usable #File(s) listed in sfm_data: 11
usable #Intrinsic(s) listed in sfm_data: 1
2. Compute features
INFO: [main_ComputeFeatures.cpp:120]  You called : 
/home/jm/Documents/openMVG_Build/Linux-x86_64-RELEASE/openMVG_main_ComputeFeatures
--input_file /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches/sfm_data.json
--outdir /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches
--describerMethod SIFT
--upright 0
--describerPreset NORMAL
--force 1
--numThreads 0

INFO: [loggerprogress.hpp:79] [- EXTRACT FEATURES -] 100%
INFO: [main_ComputeFeatures.cpp:343] Task done in (s): 1
2. Compute matches
INFO: [main_ComputeMatches.cpp:112]  You called : 
/home/jm/Documents/openMVG_Build/Linux-x86_64-RELEASE/openMVG_main_ComputeMatches
--input_file /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches/sfm_data.json
--output_file /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches/matches.putative.bin
--pair_list 
Optional parameters:
--force 1
--ratio 0.8
--nearest_matching_method ANNL2
--cache_size unlimited
--preemptive_feature_used/count 0 / 200
INFO: [loggerprogress.hpp:79] [- Regions Loading -] 100%
INFO: [main_ComputeMatches.cpp:215]  - PUTATIVE MATCHES - 
ERROR: [main_ComputeMatches.cpp:289] Invalid Nearest Neighbor method: ANNL2
2. Filter matches
INFO: [main_GeometricFilter.cpp:131]  You called : 
/home/jm/Documents/openMVG_Build/Linux-x86_64-RELEASE/openMVG_main_GeometricFilter
--input_file:        /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches/sfm_data.json
--matches:           /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches/matches.putative.bin
--output_file:       /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches/matches.f.bin
Optional parameters: 
--input_pairs        
--output_pairs       
--force              false
--geometric_model    f
--guided_matching    0
--cache_size         unlimited
INFO: [loggerprogress.hpp:79] [- Regions Loading -] 100%
ERROR: [indMatch_utils.cpp:79] Cannot open the matche file: /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches/matches.putative.bin.
ERROR: [main_GeometricFilter.cpp:256] Failed to load the initial matches file.
3. Do Incremental/Sequential reconstruction
INFO: [main_SfM.cpp:157] 
-----------------------------------------------------------
 Structure from Motion:
-----------------------------------------------------------
INFO: [loggerprogress.hpp:79] [- Features Loading -] 100%
ERROR: [main_SfM.cpp:466] Cannot load the match file.
5. Colorize Structure
ERROR: [main_ComputeSfM_DataColor.cpp:69] The input SfM_Data file "/home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/reconstruction_sequential/sfm_data.bin" cannot be read.
4. Structure from Known Poses (robust triangulation)
INFO: [main_ComputeStructureFromKnownPoses.cpp:56] Compute Structure from the provided poses
ERROR: [main_ComputeStructureFromKnownPoses.cpp:131] The input SfM_Data file "/home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/reconstruction_sequential/sfm_data.bin" cannot be read.
2. Filter matches (for the global SfM Pipeline)
INFO: [main_GeometricFilter.cpp:131]  You called : 
/home/jm/Documents/openMVG_Build/Linux-x86_64-RELEASE/openMVG_main_GeometricFilter
--input_file:        /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches/sfm_data.json
--matches:           /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches/matches.putative.bin
--output_file:       /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches/matches.e.bin
Optional parameters: 
--input_pairs        
--output_pairs       
--force              false
--geometric_model    e
--guided_matching    0
--cache_size         unlimited
INFO: [loggerprogress.hpp:79] [- Regions Loading -] 100%
ERROR: [indMatch_utils.cpp:79] Cannot open the matche file: /home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/matches/matches.putative.bin.
ERROR: [main_GeometricFilter.cpp:256] Failed to load the initial matches file.
3. Do Global reconstruction
INFO: [main_SfM.cpp:157] 
-----------------------------------------------------------
 Structure from Motion:
-----------------------------------------------------------
ERROR: [sfm_features_provider.hpp:62] Invalid feature files for the view: /home/jm/Documents/openMVG_Build/software/SfM/ImageDataset_SceauxCastle/images/100_7105.JPG
ERROR: [sfm_features_provider.hpp:62] Invalid feature files for the view: /home/jm/Documents/openMVG_Build/software/SfM/ImageDataset_SceauxCastle/images/100_7100.JPG
ERROR: [sfm_features_provider.hpp:62] Invalid feature files for the view: /home/jm/Documents/openMVG_Build/software/SfM/ImageDataset_SceauxCastle/images/100_7104.JPG
ERROR: [sfm_features_provider.hpp:62] Invalid feature files for the view: /home/jm/Documents/openMVG_Build/software/SfM/ImageDataset_SceauxCastle/images/100_7102.JPG
ERROR: [sfm_features_provider.hpp:62] Invalid feature files for the view: /home/jm/Documents/openMVG_Build/software/SfM/ImageDataset_SceauxCastle/images/100_7103.JPG
ERROR: [sfm_features_provider.hpp:62] Invalid feature files for the view: /home/jm/Documents/openMVG_Build/software/SfM/ImageDataset_SceauxCastle/images/100_7101.JPG
ERROR: [main_SfM.cpp:452] Cannot load view corresponding features in directory: .
5. Colorize Structure
ERROR: [main_ComputeSfM_DataColor.cpp:69] The input SfM_Data file "/home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/reconstruction_global/sfm_data.bin" cannot be read.
4. Structure from Known Poses (robust triangulation)
INFO: [main_ComputeStructureFromKnownPoses.cpp:56] Compute Structure from the provided poses
ERROR: [main_ComputeStructureFromKnownPoses.cpp:131] The input SfM_Data file "/home/jm/Documents/openMVG_Build/software/SfM/tutorial_out/reconstruction_global/sfm_data.bin" cannot be read.

NB: A tip for the french users : don't compile openMVG in "~/Téléchagements"! 'arrow' doesn't like the accents! That breaks the compilation!

[Ocezo]

Oups... I found the solution in the sources (as usual!)... Change the Python script "tutorial_demo.py". At the line 57 write "AUTO" vs "ANNL2" and it works. Note that openMVG_main_ComputeMatches.cpp proposes all these possibilities within the variable : "sNearestMatchingMethod". You can choose: "AUTO", "BRUTEFORCEL2", "BRUTEFORCEHAMMING", "HNSWL2", "HNSWL1", "HNSWHAMMING", "CASCADEHASHINGL2" and "FASTCASCADEHASHINGL2". Maybe some don't work or ask other arguments. I didn't check every methods.