This is an ongoing project that aims to integrate voxel hashing techniques for real-time reconstruction of dynamic scene from a single, moving RGB-D camera. The main algorithmic insipirations are KillingFusion and SobolevFusion from the dynamic scene reconstruction perspective see Mira Slavcheva's page at TUM , and work of Matthias Neissner's group on spatial voxel hashing in the context of voxel-based 3D reconstruction. The original code is forked from InfiniTAM, Active Vision Group at Oxford University, but many parts have been massively modernized -- essentially, rewritten -- and extended with my own code (hence you'll find many files with a less-restrictive Apache V2 license insted of the original Oxford University Innovation Academic License.) Having said that, I haven't tested the older static-scene algorithms in awhile, so try those at your own risk at this point.
<build_folder>/Apps/InfiniTAM/InfiniTAM --config=Files/infinitam_snoopy_config.json
Note: If you build with FFMPEG, this will also record a video in the output folder specified in the config file.
Yes, after a looong break, I'm officially switching to try to do something with it again, at least for awhile. Even if I'm not actively working on it, I do my best to respond to new issues or collaboration requests.
TLDR: Maybe.
Originally, that was the plan. However, at the time of writing I'm exhausted from trying to stick to the open(for extension)-closed(for modification) principle. The code wasn't originally designed for dynamic-scene fusion, so some things really do need to change to make the new things more maintainable. I've already changed a lot of code from the original InfiniTAM codebase, IMHO, for the better.
If this fork achieves reasonable success, I'll reach out to the InfiniTAM authors and ask whether they'd like to work on integrating my changes into their codebase. I (hope that I) didn't break anything so far, but the original code lacks any hint of continuous integration, which I plan to add to it, at least for my code, shortly.
Many portions of this code are still privy to the original Oxford University Innovation Academic License included within, consult file headers for which license is used for which part of the code.
Original InfiniTAM Repo & README: https://github.com/victorprad/InfiniTAM
此处可能存在不合适展示的内容,页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容,可点击提交进行申诉,我们将尽快为您处理。