YujiaoShi / Boosting3DoFAccuracy

ICCV2023: Boosting 3-DoF Ground-to-Satellite Camera Localization Accuracy via Geometry-Guided Cross-View Transformer
MIT License
41 stars 4 forks source link

The results obtained using the pre-trained model were not good #6

Closed 919106840533 closed 1 month ago

919106840533 commented 1 month ago

Thank you for your research and sharing. When I used the pre-trained model to test on Ford Log2, the results I got were quite different from those shown by the pre-trained model. The parameters I used in the test were all default parameters in the code, and I wonder if it is the parameter setting problem. The following are the test results of the pre-trained model given in Ford Log2 and the results obtained by me using the pre-trained model. GPU used in the experiment: 2080TI results I got: Screenshot from 2024-10-01 16-00-31

results you shown: 1

919106840533 commented 1 month ago

Problem solved