FLM-Occ: Feed-forward Likelihood Maximization for Efficient Indoor Occupancy Prediction

2026-06-19

AI Summary

Existing indoor occupancy prediction methods rely on voxel-wise classification and lack global supervision over Gaussian primitive distributions, often generating spurious primitives in empty regions, which degrades both representational fidelity and computational efficiency. This work reframes occupancy prediction as a voxel distribution estimation problem and introduces the Feedforward Likelihood Maximization (FLM) framework. FLM employs mixture density modeling to predict primitive distributions and directly maximizes the likelihood of observed occupied voxels during feedforward inference, enabling end-to-end training. The method innovatively defines mixture weights as normalized primitive volumes to implicitly satisfy simplex constraints and derives a differentiable voxelization formula compatible with standard mixture models, facilitating long-range primitive relocation. On Occ-ScanNet, the approach achieves higher accuracy using only 32 superquadrics—just 2.7% of the primitives used by the previous state-of-the-art—while accelerating inference by 3.7×.

Abstract

Recent indoor occupancy prediction methods adopt Gaussian primitives as a sparse 3D representation for computational efficiency. However, their training relies on voxel classification, which imposes only local constraints and lacks global supervision on the distribution of the primitives. Therefore, they inevitably predict spurious primitives in empty regions, undermining both representational and computational efficiency. To address this, we propose Feed-forward Likelihood Maximization (FLM), a novel framework that reformulates occupancy prediction as voxel distribution estimation. In FLM, a network is trained to predict a mixture model that maximizes the likelihood over ground-truth occupied voxels in a feed-forward manner. To enable end-to-end training of networks and voxelization of a standard mixture model, we define mixture weights as normalized primitive volumes to implicitly enforce simplex constraints and derive novel voxelization formulas. Based on FLM, our FLM-Occ, a novel method that is capable of relocating randomly initialized primitives over long distances to model a scene. On Occ-ScanNet, FLM-Occ achieves superior accuracy using only 32 superquadrics, 2.7% of the prior SoTA, while running 3.7 times faster.

Problem

• indoor occupancy prediction
• Gaussian primitives
• voxel classification
• global supervision
• spurious primitives

Innovation

• Feed-forward Likelihood Maximization
• occupancy prediction
• mixture model
• voxelization
• superquadrics