search

WU-CVGL / MBA-SLAM

"MBA-SLAM: Motion Blur Aware Dense Visual SLAM with Radiance Fields Representation"
https://arxiv.org/abs/2411.08279
51 stars 1 forks source link

Input sensors #1

Open cdcseacave opened 1 week ago

cdcseacave commented 1 week ago

Great paper, and amazing results. Did you also test how your algorithm works with lower quality RGBD inputs, like from an iPhone Pro?

wangpeng000 commented 1 week ago

Thank you for your prompt attention to our work! Unfortunately, we do not have access to an iPhone Pro device. However, we tested MBA-SLAM using real sequences captured by a RealSense camera. Although the RGB and depth data quality from the RealSense camera is not ideal, MBA-SLAM performed well.

cdcseacave commented 1 week ago

Indeed, but I'm afraid the quality of iPhone RGBD images is even worse. May I send you a dataset to test?

wangpeng000 commented 1 week ago

@cdcseacave Of course, I’ll test your dataset this week. My email: wangpeng AT westlake.edu.cn

cdcseacave commented 1 week ago

pls try this: https://filetransfer.io/data-package/tLVLxsVS#link

wangpeng000 commented 5 days ago

@cdcseacave Hi, we are currently processing your data due to an unmatched resolution, which will take some time. Additionally, the low-resolution depth data captured by the iPhone is really not so good, so the results may not turn out as expected.

cdcseacave commented 5 days ago

Indeed, that is why I was afraid too. Thank you for trying, looking forward to the results.