Decision-FLIP: Budget-Free Structured Pruning for Gaussian Occupancy Models

Jingxing Li; Anurag Paul; Subhasis Das; Cindy Xu

cs.CVOct 14, 2026

Decision-FLIP: Budget-Free Structured Pruning for Gaussian Occupancy Models

Authors: Jingxing Li, Anurag Paul, Subhasis Das, Cindy Xu

Organizations: Zoox Inc.

Abstract

Occupancy predictors rely on large persistent banks of Gaussians or learned queries whose size is fixed by hand. Common pruning proxies, such as opacity, activation frequency, or parameter magnitude, do not measure whether removing a row changes the model's voxel decisions. We introduce Decision-FLIP, a budget-free compilation framework whose scoring and structure selection need only unlabeled calibration inputs. Exact Mean-FLIP scores each persistent row by the number of voxel argmax decisions that change when the row is removed from the readout; a guarded Otsu split of the log-score distribution selects the survivor count without a target and abstains when the split is unsupported; and the survivors are physically compiled into a smaller checkpoint, so camera sampling, attention, and aggregation shrink with the bank. For streaming banks, scores are integrated over each row's lifecycle through warp, death, and refill. Across four Gaussian-bank model-dataset settings and one persistent-query transfer, the automatically selected cuts remove 12-79% of eligible rows and cut bank/query-dependent latency by 7-45% across settings with component profiles, with whole-model savings of up to 8.0% on the primary endpoints bounded by fixed-cost image, voxel, and sensor modules and peak-memory savings of up to 16.6%. Under matched one-epoch recovery controls, semantic mIoU changes by at most 0.18 point. Same- $K$ controls show that the ranking, not only the bank size, matters: random removal collapses GaussianFormer-1 and OccFormer, while label-free Mean-FLIP matches or outperforms supervised Taylor ranking on all five settings where that control was run. The method applies when removable primitives keep a stable checkpoint-row identity; it is budget-free in survivor count, not parameter-free.

Figures & tables

Figure 1: Overview of Decision-FLIP. (a) On unlabeled calibration frames, removing row

i

from the cached full readout and recomputing the all-class argmax gives per-frame Decision-FLIP; scores aggregate as Mean-FLIP or lifecycle Mean-FLIP. (b) Guarded log-Otsu selects a survivor set only when

\eta\geq 0.5

, otherwise retaining the full bank. (c) The same survivor indices are gathered across aligned tensors, strict-loaded, and compared with the unpruned reference under matched one-epoch recovery and official evaluation.

Model / metrics	Kept / total (removed)	Before $\Delta$	After recovery $\Delta$	Bank/query latency $\Delta$	End-to-end latency $\Delta$	Peak memory $\Delta$
GaussianWorld, primary lifecycle compilation (mIoU / geometry IoU)	22,427 / 25,600 (12.4%)	-0.179 / -0.290	-0.082 / -0.119	-7.0%	-2.4%	-1.2%
GaussianWorld, non-primary static-frame reference (mIoU / geometry IoU)	16,461 / 25,600 (35.7%)	-0.412 / -0.574	-0.225 / -0.270	-24.3%	-9.3%	0.0%
GaussianFormer-1 (mIoU / geometry IoU)	17,365 / 25,600 (32.2%)	-0.157 / -0.435	-0.171 / -0.246	-23.3%	-5.5%	+0.8%
GaussianFormer3D (mIoU / geometry IoU)	22,535 / 25,600 (12.0%)	-0.044 / -0.093	-0.024 / -0.116	-13.2%	-4.7%	-8.3%
GaussianFlowOcc (voxel mIoU / occupied IoU)	8,628 / 10,000 (13.72%)	+0.10 / +0.57	+0.01 / +0.03	-10.7%	-8.0%	-7.0%
OccFormer transfer, SemanticKITTI (SSC mIoU / SC IoU)	21 / 100 (79.0%)	-0.16 / 0.00	+0.03 / +0.08	-45.0%	-5.7%	-16.6%

Table 1: Automatic compilation across model-dataset settings. Changes are compiled minus full in the displayed metric order.

Figure 2: Qualitative preservation under lifecycle-aware GaussianWorld compilation. The two validation samples (scene indices 37 and 112, frame 19) were fixed before inspecting predictions. Each model processes the same 20-frame streaming prefix after matched recovery. Decision-FLIP retains 22,427 of 25,600 rows (12.4% removed). Difference panels separate full occupied to compiled empty, full empty to compiled occupied, and semantic changes where both predictions remain occupied; percentages use all 640,000 voxels in each 200 $\times$ 200 $\times$ 16 grid.

Figure 3: Automatic selector comparison. (a) Automatically removed fraction. (b) Zero-recovery semantic mIoU change from each model’s full source. Concentric markers indicate identical masks; arrows annotate clipped Kneedle failures. Selection never uses validation metrics.

Model	$K$	Mean-FLIP	Taylor	Random	Activity	Opacity	Magnitude
GaussianWorld	22,427	-0.179	-0.194	-0.390	-0.252	-0.244	-0.326
GaussianFormer-1	17,365	-0.157	-1.221	-8.310	-8.652	-0.183	-0.474
GaussianFormer3D	22,535	-0.044	-0.075	-0.730	-0.100	-0.105	-0.071
GaussianFlowOcc	8,628	+0.10	+0.10	n/a	n/a	n/a	n/a
OccFormer (SemanticKITTI transfer)	21	-0.16	-7.98	-12.68	-0.33	n/a	-11.97

Table 2: Same-

K

ranking-criterion ablation before recovery (primary semantic metric

\Delta

vs. full). We report one primary automatically selected operating point per architecture; the non-primary GaussianWorld static reference is excluded.

Appendix figures & tables4 assets

Supplementary material from the paper’s appendix.

Appendix

Figure A1: Cross-architecture SurroundOcc qualitative results. Rows show the primary endpoints after matched recovery for GaussianFormer-1 (full $K=25{,}600$ , compiled $K=17{,}365$ ) and GaussianFormer3D (full $K=25{,}600$ , compiled $K=22{,}535$ ). Columns are ground truth (GT), Full, Decision-FLIP Compiled, and Difference. Difference percentages use all 640,000 voxels in each 200 $\times$ 200 $\times$ 16 grid as the denominator.

Figure A2: Cross-architecture qualitative results for the exact primary matched-recovery endpoints: OccFormer on SemanticKITTI (full $K=100$ , compiled $K=21$ ) and GaussianFlowOcc on Occ3D-nuScenes ( Tian et al., 2023 ) (full $K=10{,}000$ , guarded-Otsu compiled $K=8{,}628$ , $\tau=0.2$ ). Both samples were fixed before inspecting model output. Columns are ground truth (GT), Full, Decision-FLIP Compiled, and Difference. OccFormer percentages use all 2,097,152 voxels in its 256 $\times$ 256 $\times$ 32 grid; GaussianFlowOcc percentages use the 130,720 camera-visible voxels in its 200 $\times$ 200 $\times$ 16 grid.

Model	Largest class drop	Largest upward change
GaussianWorld	vegetation -0.26	trailer +0.15
GaussianFormer-1	vegetation -0.40	none (best: construction vehicle -0.02)
GaussianFormer3D	manmade -0.18	other flat +0.16
OccFormer (paired evaluation)	other-vehicle -0.59	terrain +0.72
GaussianFlowOcc	trailer -0.33	barrier +0.32

Appendix

Table A9: Per-class preservation across architectures

Figure A3: Detailed guarded-Otsu diagnostic on GaussianFormer-1. Left: the saved log Mean-FLIP distribution and selected partition. Right: normalized between-group variance over candidate thresholds; guarded Otsu compiles only when the maximizing separation exceeds the fixed gate. This diagnostic is retained as distributional evidence and does not imply universal bimodality or accuracy optimality.

Model Compression

Model compression reduces the storage, computation, or latency of machine-learning models while preserving task performance.

See all 122 papers

Oct 11, 2026

GNFormer: Gaussian Neural Field Transformer for Efficient Camera-based Occupancy Prediction

Camera-based occupancy prediction provides geometric and semantic scene understanding for robot perception and has been adopted in autonomous vehicle and UAV perception systems. Gaussian primitives offer a promising alternative by potentially representing only occupied regions, thereby concentrating computation and representational capacity where needed and reducing wasted computation and memory in empty space. However, existing Gaussian-based methods suffer from poor primitive utilization, as ineffective training objectives leave some Gaussians in empty space, while the limited expressiveness of Gaussian functions restricts their ability to model complex shapes. To address these limitations, we propose a novel 3D representation, Gaussian Neural Field (GNF), which can accurately map voxel positions to semantic occupancies. In GNF, Gaussians and a neural network jointly parameterize the field, where the Gaussians define its spatial support and provide features for neural decoding, while the neural network enables more expressive occupancy modeling than Gaussian functions alone. We further develop the Gaussian Neural Field Transformer (GNFormer), a streamlined and flexible method that predicts GNFs for camera-based occupancy prediction. Across Occ-ScanNet, SSCBench-KITTI-360, and nuScenes-SurroundOcc benchmarks, GNFormer achieves the highest IoU and mIoU and the fastest inference among the compared single-frame methods. In the experiments, GNFormer can represent indoor and outdoor scenes with only 256 and 1024 Gaussians, respectively, substantially fewer primitives than prior Gaussian-based methods. The code will be released.

Guangcheng Chen, Li He, Hong Zhang

Department of Electronic and Electrical Engineering, Southern University of Science and Technology, China · Shenzhen Key Laboratory of Robotics and Computer Vision, Southern University of Science and Technology, Shenzhen, China

3D Occupancy Prediction 3D ViTs

Ask PDF

Oct 2, 2026

Budgeted-GS: Real-Time Large-Scale Gaussian Splatting via Factoring LOD

3D Gaussian Splatting achieves excellent visual quality with real-time rendering, but at the scale of entire cities it does not fit: a trained model carries millions of primitives and gigabytes of memory, and real-time rendering at high quality on a consumer GPU remains out of reach. We introduce Budgeted-GS, a post-hoc method that turns any trained 3DGS model into a factoring tree, a multi-resolution hierarchy of moment-matched aggregates. After a construction pass of a few seconds, a single quality parameter selects, for each view, the level of detail that fits the memory of the target device, so the same city-scale model serves GPUs with widely different memory capacities. When a new scene is to be trained, the same theory applies: instead of growing a full-sized model and compressing it afterwards, budget-centered training first measures how many primitives the scene needs and then trains the model directly at that size, avoiding the wasted effort of optimizing primitives that are later discarded. Both methods are grounded in a measurable capacity floor, a budget-error law derived from optimal transport in phase space; selection rules certified by recent covering theorems decide which primitives are redundant. The floor answers how many primitives a scene actually needs and how many can safely be given up. We validate the floor on 13 public scenes under a preregistered protocol, and exercise both methods from object scenes to an official city capture, rendering it at native 1920x1080, full SH, in real time on one consumer GPU.

Haipeng Wang

Neusoft

3DGS Training 3D Gaussian Splatting 3DGS Acceleration 3DGS Compression

Ask PDF

Jun 19, 2026

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

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.

Guangcheng Chen, Lihuang Fang, Huaqi Tao +3

Southern University of Science and Technology · Shenzhen Key Laboratory of Robotics and Computer Vision · Guangdong Laboratory of Artificial Intelligence and Digital Economy (Shenzhen)

Superquadric Representation 3D Occupancy Prediction

Ask PDF

Abstract

Model Compression

Explore similar work

GNFormer: Gaussian Neural Field Transformer for Efficient Camera-based Occupancy Prediction

Budgeted-GS: Real-Time Large-Scale Gaussian Splatting via Factoring LOD

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