@Almorgand Thank you for posting our work!

@mervenoyann Thank you! We have already built a demo on HF (~

@edgarriba @ducha_aiki Hi Edgar, thanks for your interest in our work! We are glad that our model can be considered included in kornia. The models and code will be open-sourced later. We can convert them (our pretrained ELoFTR and ROMA models) to onnx if needed.

@XylonSean Hi Xuelun, thanks for your response! We clarify that the comparison with GIM in this section highlights how merging detector-free matches with a large radius and refining them produces longer tracks. We’ll revise the comment on GIM's propagation to distant frames in revision.

@ducha_aiki @vbalnt Thanks for the splendid challenge! I'm curious about how to combine three metrics to determine the winner~