Yuchen Song (@YuchenEasonSong)
2025-03-21 | โค๏ธ 10 | ๐ 3
Considering embedding knowledge from foundation models into 3D Gaussian Splatting? Check out M3, our novel approach achieving superior performance over distillation-based methods through efficient feature alignment. ICLR2025
๐ Related
Auto-generated - needs manual review
์ธ์ฉ ํธ์
Xueyan Zou (@xyz2maureen)
[1/n] We are releasing M3 (#ICLR2025): a Gaussian Splatting method that builds LMM memories for arbitrary scenes. ๐ฅ [Efficient] 16 degrees in each Gaussian primitive for one LMM. ๐ฅ [Alignment] The rendered features are directly in the source LMM embedding space. https://t.co/PGOmhAzzWs
๐ฌ ์์
