Closed projan-shakya closed 2 weeks ago
Hi, would it be possible for you to share one of these images? Could you please let us know your OS?
im running on windows 11
This might be a COLMAP issue. Did all the steps before COLMAP's hierarchical_mapper in generate_colmap.py
run smoothly?
oh im sorry!! This error was on generate_colmap.py i mistyped to .py file my bad
Could you please run generate_colmap.py and check that feature_extractor
, make_colmap_custom_matcher.py
and matches_importer
run smoothly? running generate_colmap.py with a debugger might help
i changed the dataset from .png to .jpg and the file seem to have created colmap file but while running train_full.py it showed that --alpha_masks/-a expected one argument is it necessary to create masked images?
the full error
usage: train_coarse.py [-h] [--sh_degree SH_DEGREE] [--source_path SOURCE_PATH] [--model_path MODEL_PATH]
[--exp_name EXP_NAME] [--images IMAGES] [--alpha_masks ALPHA_MASKS] [--depths DEPTHS]
[--resolution RESOLUTION] [--white_background] [--train_test_exp]
[--data_device DATA_DEVICE] [--eval] [--skip_scale_big_gauss] [--hierarchy HIERARCHY]
[--pretrained PRETRAINED] [--skybox_num SKYBOX_NUM] [--scaffold_file SCAFFOLD_FILE]
[--bounds_file BOUNDS_FILE] [--skybox_locked] [--iterations ITERATIONS]
[--position_lr_init POSITION_LR_INIT] [--position_lr_final POSITION_LR_FINAL]
[--position_lr_delay_mult POSITION_LR_DELAY_MULT]
[--position_lr_max_steps POSITION_LR_MAX_STEPS] [--feature_lr FEATURE_LR]
[--opacity_lr OPACITY_LR] [--scaling_lr SCALING_LR] [--rotation_lr ROTATION_LR]
[--exposure_lr_init EXPOSURE_LR_INIT] [--exposure_lr_final EXPOSURE_LR_FINAL]
[--exposure_lr_delay_steps EXPOSURE_LR_DELAY_STEPS]
[--exposure_lr_delay_mult EXPOSURE_LR_DELAY_MULT] [--percent_dense PERCENT_DENSE]
[--lambda_dssim LAMBDA_DSSIM] [--densification_interval DENSIFICATION_INTERVAL]
[--opacity_reset_interval OPACITY_RESET_INTERVAL] [--densify_from_iter DENSIFY_FROM_ITER]
[--densify_until_iter DENSIFY_UNTIL_ITER]
[--densify_grad_threshold DENSIFY_GRAD_THRESHOLD]
[--depth_l1_weight_init DEPTH_L1_WEIGHT_INIT]
[--depth_l1_weight_final DEPTH_L1_WEIGHT_FINAL] [--convert_SHs_python]
[--compute_cov3D_python] [--debug] [--ip IP] [--port PORT] [--debug_from DEBUG_FROM]
[--detect_anomaly] [--save_iterations SAVE_ITERATIONS [SAVE_ITERATIONS ...]] [--quiet]
[--checkpoint_iterations CHECKPOINT_ITERATIONS [CHECKPOINT_ITERATIONS ...]]
[--start_checkpoint START_CHECKPOINT]
train_coarse.py: error: argument --alpha_masks/-a: expected one argument
Error executing train_coarse: Command 'python train_coarse.py -s try\camera_calibration\aligned --save_iterations -1 -i ../rectified/images --skybox_num 100000 --model_path try\output\scaffold --alpha_masks ' returned non-zero exit status 2.```
Hi, I'm glad you managed to calibrate It is a bug, thanks for spotting it. I just pushed a fix. Let me know how it goes
Thanks for the quick update, the training has started Will update on how it goes
Hello @ameuleman sorry for the late response
The masking error still exists and it showed this error after training started for a bit
I made a mistake when in the full_train script. It should be fixed now. To avoid retraining coarse, you can pass --skip_if_exists
.
It works perfectly fine for me now thank you. However it seems like for Colmap generation we need large dataset of image since it is hierarchical mapper.
Hi, Our method focuses on large scene reconstruction (tens of thousands of images). We designed our calibration scripts with this kind datasets in mind
Error executing colmap hierarchical_mapper: Command '['colmap.bat', 'hierarchical_mapper', '--database_path', python preprocess/generate_colmap.py --project_dir E:\2025GStest001\fengcheData
Error executing colmap hierarchical_mapper: Command '['colmap.bat', 'hierarchical_mapper', '--database_path', python preprocess/generate_colmap.py --project_dir
When i run the code with the path of the dataset being the example_dataset u provided it works completely fine
but when i run it with my dataset with the folder structure same as that of example_dataset it shows this error
this error goes long end terminates with this text