SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2404.00815 · CVPR · 2024

Towards Realistic Scene Generation with LiDAR Diffusion Models

Vitor Guizilini, Yue Wang, Haoxi Ran

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 8 functions out of this paper's own repositories and ran 5 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
hancyran/lidar-diffusion — 5 of 8
FunctionStatusWhere it lives
AttnBlock Ran hancyran/lidar-diffusion/lidm/modules/diffusion/model_lidm.py
code served (permissive licence) · get_code("d1c55266c7801e2b")
CircularConv2d Ran hancyran/lidar-diffusion/lidm/modules/diffusion/model_lidm.py
code served (permissive licence) · get_code("99317db8781524bf")
Downsample Ran hancyran/lidar-diffusion/lidm/modules/diffusion/model_lidm.py
code served (permissive licence) · get_code("ba09c4650052a805")
LinAttnBlock Ran hancyran/lidar-diffusion/lidm/modules/diffusion/model_lidm.py
code served (permissive licence) · get_code("1be48aa5423551e2")
LinearAttention Ran hancyran/lidar-diffusion/lidm/modules/diffusion/model_lidm.py
code served (permissive licence) · get_code("65eaa953ffaffaad")
Encoder Not yet run hancyran/lidar-diffusion/lidm/modules/diffusion/model_lidm.py
code served (permissive licence) · get_code("6b0bf22688da1927")
ResnetBlock Not yet run hancyran/lidar-diffusion/lidm/modules/diffusion/model_lidm.py
code served (permissive licence) · get_code("c54f70d846400a40")
make_attn Not yet run hancyran/lidar-diffusion/lidm/modules/diffusion/model_lidm.py
code served (permissive licence) · get_code("321eebc8c4fd6b54")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Curve-based DMs (Ours) Throughput↑: 1.603 samples/sec (×107) Patch-based DMs (Latent Diffusion) Throughput↑: 2.171 samples/sec (×145) Point-based DMs (LiDARGen) Throughput↑: 0.015 samples/sec (×1) Figure 1. Our method (LiDM) steps towards LiDAR-realistic scene generation by preserving curve-like structures and objects with greater resemblance to real-world data (Reference), and marks a milestone for conditional LiDAR scene generation from different input modalities.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2404.00815")
get_code_for_paper("2404.00815")
have("2404.00815")

Connect an agent — have() is free.