Listed minimum requirements vs. paper?

[olov-andersson]

Thanks for the nice work! I am just wondering what the minimum requirements actually are for a room-sized environment

The paper says it runs in real-time on a 3060M (mobile) and as far as I know those usually come with only 6GB of VRAM. The repo instructions claims it needs 10+ GB but also seems to indicate that >5GB might work if the environment TSDF is small? Is there some tuning needed of the TSDF part to get this to work in a room-sized environment?

[Kashu7100]

Thank you for your interest in our project.

We have integrated Open3D for its CUDA-accelerated TSDF (Truncated Signed Distance Function) capabilities. It's important to note that this particular TSDF implementation dynamically adjusts GPU memory usage based on the scene's dimensions, meaning it allocates additional GPU resources as the scene expands. Our SEEM framework typically requires a constant 3-5 GB of GPU memory, though this can vary with the scene size and specific configurations.

To mitigate the memory usage, consider lowering the voxel resolution. This adjustment can significantly reduce memory demands.

Regarding larger environments, such as full-room scans, a GPU with more substantial memory becomes necessary. Alternatively, offloading computations to an external server could be a viable solution. We recognize this limitation and are actively working on optimizing our system to lower GPU requirements. Our goal is to release an updated version of the code concurrent with the acceptance of our upcoming paper, which will include these improvements.