KP2D
KP2D copied to clipboard
TRI-ML
Hi, First of all, thank you for this awesome project. When I run "make docker-build" command, I encounter the following error: E: Unable to locate package libcudnn7 E: Unable to...
Thank you for sharing the code, but when I configure the environment to run the code, loss will become Nan in a short time. I guess the problem may be...
From [here](https://github.com/TRI-ML/KP2D/blob/77633af57d65c5a0cd7609eaeab7e6cb758ca791/kp2d/datasets/augmentations.py#L344) we know that the source image is the result of homography H applied to target image. We can also know that the source information in model part is...
 What model did you use for V0, V1, V2 after submission? keypoint_net or keypoint_resnet?
Hello, I am attempting to run the model on higher resolution(1024x768), but I got CUDA out of memory error on a 12gb GPU. If I downscale the images to half...
Thank you for the great work! Looking into the `evaluate_keypoint_net` function I can see that you use the `to_color_normalized` function to apparently normalize the input images. However this function does:...
Hello, Many thanks for releasing the code. I'm wondering if the code actually support resuming from a checkpoint file? It looks like the function `parse_train_file` in kp2d/utils/config.py rejects a .ckpt...
Hi again, I looking for a more adapted metric for my use and wanted to know if there is a unit for **mean_dist** value in **compute_homography** function in descriptor_evaluation.py. If...
Thanks for the sharing. I would like to train a kp2d model with a large rotation angle, e.g., pi. However, I find the recall decreases in such training setting, although...
I was wondering if it is possible to build the repo without docker image since it limits the use, in particular windows based machines tends to have major issues with...
