You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello, can you release the code about extracting and matching R2D2+PUMP feature points on HPatches dataset?
Now, I have implemented R2D2+PUMP according to extract.py released in R2D2. In my implementation, I detect multi-scale R2D2 keypoints and corresponding PUMP descriptors. The adopted R2D2 model is r2d2_WASF_N16.pt. The adopted PUMP model is PUMP.pt. Then I use the evaluation code released in D2-Net to compute the MMA@3 of R2D2+PUMP.
But, the MMA@3 I got is only 48.67, which is much lower than 84.16, the MMA@3 reported in the PUMP paper.
The text was updated successfully, but these errors were encountered:
Hello, can you release the code about extracting and matching R2D2+PUMP feature points on HPatches dataset?
Now, I have implemented R2D2+PUMP according to extract.py released in R2D2. In my implementation, I detect multi-scale R2D2 keypoints and corresponding PUMP descriptors. The adopted R2D2 model is r2d2_WASF_N16.pt. The adopted PUMP model is PUMP.pt. Then I use the evaluation code released in D2-Net to compute the MMA@3 of R2D2+PUMP.
But, the MMA@3 I got is only 48.67, which is much lower than 84.16, the MMA@3 reported in the PUMP paper.
The text was updated successfully, but these errors were encountered: