Closed Charlotte0592 closed 2 years ago
Can you provide more details on what you are trying to do? How do you obtain the features?
According to Aachen v1.1 dataset, I wrote a images_list.TXT by myself and then used extract.py in R2D2 to extract local features, using the following command:
${PYTHONBIN} extract.py --model my_models/${LOCAL}.pt --images ${DATASETS_PATH}/images_list.txt --min-scale 0.3 --min-size 0 --max-size 9999 --top-k ${KPTS} --tag ${METHOD}
And I want to check whether my algorithm to extract local features is work, so I run
python reconstruction_pipeline_aachen_v1_1.py \
--dataset_path ${DATASETS_PATH} \
--colmap_path colmap \
--method_name ${METHOD}
I want to use only local features to achieve visual localization, so I run this pipeline to obtain the queries' pose, in order to check whether my own algorithm is good. But I ran into the problem as described above. Could you please give some advice?
The error sqlite3.IntegrityError: UNIQUE constraint failed: keypoints.image_id
indicates that there are two images with the same image_id. I haven't seen this happening before though. Did you change anything else?
Thanks for sharing the amaing benchmark! I have some promblems when I run:
then, it happens:
Does it means there are two or more same keypoints? Could please help me solve this problem?