CMU dataset about hfnet HOT 9 CLOSED

Sorry for the late reply. This repo provides evaluation for the CMU Seasons dataset, not the extended version. We used this script to generate a SIFT database from the .nvm file. It is called from the reconstruction script. I think that there is no sift_queries.db file, this must be a typo in the documentation.

from hfnet.

Thank you for your reply. I used the magic_cmu_to_db.py script to generate the sift_database.db for each slice, and found the create_cmu_query_db.py script for the sift_queries.db (not sure if it's needed like you said, but I generated them anyway). Now trying to run the evaluation, hopefully that works!

from hfnet.

When running evaluate_cmu.py, I get the following error, which seems to be related to PCA:

ValueError: n_components=1024 must be between 0 and min(n_samples, n_features)=446 with svd_solver='full'

The min(n_samples, n_features) number changes depending on the slice. Have you encountered this, or know how to fix the issue? I saw in another issue that you had mentioned possibly disabling PCA?

from hfnet.

When running evaluate_cmu.py, I get the following error, which seems to be related to PCA:

ValueError: n_components=1024 must be between 0 and min(n_samples, n_features)=446 with svd_solver='full'

The min(n_samples, n_features) number changes depending on the slice. Have you encountered this, or know how to fix the issue? I saw in another issue that you had mentioned possibly disabling PCA?

I've just met the same problem with you, have you solved it now?

from hfnet.

When running evaluate_cmu.py, I get the following error, which seems to be related to PCA:
ValueError: n_components=1024 must be between 0 and min(n_samples, n_features)=446 with svd_solver='full'
The min(n_samples, n_features) number changes depending on the slice. Have you encountered this, or know how to fix the issue? I saw in another issue that you had mentioned possibly disabling PCA?

I've just met the same problem with you, have you solved it now?

Hi SheldonHS, so sorry, I didn't end up solving this because it wasn't critical for me and moved on to now trying to run hfnet on a custom dataset. Let me know if you solved it though!

from hfnet.

When running evaluate_cmu.py, I get the following error, which seems to be related to PCA:
ValueError: n_components=1024 must be between 0 and min(n_samples, n_features)=446 with svd_solver='full'
The min(n_samples, n_features) number changes depending on the slice. Have you encountered this, or know how to fix the issue? I saw in another issue that you had mentioned possibly disabling PCA?

I've just met the same problem with you, have you solved it now?

Hi SheldonHS，Is it resolved？ I also encountered the same problem

from hfnet.

When running evaluate_cmu.py, I get the following error, which seems to be related to PCA:
ValueError: n_components=1024 must be between 0 and min(n_samples, n_features)=446 with svd_solver='full'
The min(n_samples, n_features) number changes depending on the slice. Have you encountered this, or know how to fix the issue? I saw in another issue that you had mentioned possibly disabling PCA?

I've just met the same problem with you, have you solved it now?

Hi SheldonHS，Is it resolved？ I also encountered the same problem

Hi @SheldonHS and @bxh1,

Not sure if you've solved already, I just found that disabling the PCA does in fact allow the evaluation to run! In line 30 of evaluate_cmu.py I set pca_dim to 0 (instead of 1024).

from hfnet.

Hi! After set pca_dim to 0. I got another error. Could anyone help me? Thanks!
Traceback (most recent call last): File "colmap-helpers/magic_cmu_to_db.py", line 68, in <module> main() File "colmap-helpers/magic_cmu_to_db.py", line 37, in main image_id = db.add_image(name, camera_id) File "/dockerdata/xinkong/hfnet/colmap-helpers/internal/db_handling.py", line 178, in add_image prior_q[3], prior_t[0], prior_t[1], prior_t[2])) sqlite3.IntegrityError: UNIQUE constraint failed: images.name

from hfnet.

This is probably a different issue. It says that the entry in the COLMAP database already existed. Maybe the script crashed and you launched it again without deleting the corresponding db?

from hfnet.