While recent feed-forward 3D Gaussian Splatting methods enable efficient driving-scene reconstruction from surround-view images, their reliance on a single observation fundamentally hinders novel-view synthesis. Regions occluded in the input can become visible from novel viewpoints, leaving the reconstructed Gaussians under-constrained and incomplete. Moving beyond image-space inpainting for rendered views, we introduce OccSplat, a feed-forward framework whose core contribution is Occlusion-Aware Gaussian Densification, explicitly targeting missing scene geometry rather than missing pixels. By predicting new Gaussian primitives for unsupported regions, OccSplat completes the scene directly in the 3D representation. Specifically, OccSplat first identifies occluded regions that lack Gaussian support through Gaussian Occlusion Mapping. Gaussian Occlusion Completion then generates new primitives for these regions. The predicted primitives are integrated into the original representation to improve novel-view synthesis while retaining feed-forward inference without per-scene optimization. Experiments on the nuScenes benchmark demonstrate that OccSplat improves on the compared methods and yields particularly large gains in occluded regions, leading to more accurate and visually coherent driving-scene reconstruction.
Figures & tables
Figure 1: The disocclusion problem in forward-driving single-frame (SF) reconstruction. Six surround-view cameras at a source frame t (top) and target frame t+1 (bottom). As the ego-vehicle advances, content hidden behind foreground structures at t becomes disoccluded at t+1 (red dashed ellipses). A model that only observes frame t has no direct visual evidence for these regions.
Figure 2: Overall architecture of OccSplat. (a) From surround-view images {Iit} , a backbone predicts the pixel-aligned Gaussian set S0={Gi(u,v)} , after which Gaussian Occlusion Mapping (c) locates unsupported regions by rasterizing S0 against dark and bright backgrounds and using their responses to identify unsupported regions. (b) Guided by this map, Gaussian Occlusion Completion rasterizes S0 to the future view, inpaints it with an encoder-decoder, and lifts the result back into new primitives {Gocc} that densify S0 for rendering the future frame {Iit+1} .
Method
Venue
PSNR ↑
SSIM ↑
LPIPS ↓
SelfOcc [ 9 ]
CVPR 2024
18.22
0.464
–
UniPad [ 29 ]
CVPR 2024
16.45
0.375
–
EmerNeRF [ 30 ]
ICLR 2024
20.95
0.585
–
DistillNeRF [ 25 ]
NeurIPS 2024
20.78
0.590
–
MVSplat [ 3 ]
ECCV 2024
20.23
0.479
0.366
DrivingForward [ 24 ]
AAAI 2025
21.76
0.767
0.194
Table 1: Quantitative comparison of our method (OccSplat) against prior reconstruction methods on the nuScenes validation set. Best in bold, second best underlined.
Config.
Densification
Refiner
Occ. Mapping
PSNR ↑
SSIM ↑
LPIPS ↓
(a)
21.76
0.767
0.194
(b)
✓
19.92
0.747
0.182
(c)
✓
✓
20.71
0.753
0.175
(d)
✓
✓
22.27
0.765
0.187
(e)
✓
✓
✓
23.50
0.767
0.182
Table 2: Ablation study of the proposed components on the nuScenes validation set. We report PSNR ( ↑ ), SSIM ( ↑ ), and LPIPS ( ↓ ). “Densification” denotes Gaussian Occlusion Completion, “Refiner” the Pixel Refiner, and “Occ. Mapping” the dual-background Gaussian Occlusion Mapping.
ϵ
PSNR ↑
SSIM ↑
LPIPS ↓
0.99
22.91
0.764
0.187
0.8
23.41
0.767
0.183
0.7
23.50
0.767
0.182
0.6
23.30
0.757
0.192
0.5
22.55
0.741
0.204
Table 3: Effect of the occlusion threshold ϵ in Gaussian Occlusion Mapping.
Method
PSNR ↑
SSIM ↑
LPIPS ↓
DrivingForward [ 24 ]
11.498
0.759
1.178
MVSplat [ 3 ]
18.606
0.509
0.620
OccSplat (Ours)
18.602
0.763
0.465
Table 4: Performance on disoccluded regions.
Method
FPS ↑
Time ↓
VRAM (GB) ↓
Params (M) ↓
3DGS
–
∼ 9 min
–
–
DrivingForward
1.74
574.7 ms
2.96
61.88
OccSplat
1.01
992.9 ms
3.10
63.16
Table 5: Computational costs. Time denotes per-scene optimization for 3DGS and inference for the feed-forward methods.
Figure 3: Qualitative comparison on the nuScenes validation set. From top to bottom: Ground Truth, OccSplat (ours), DrivingForward, and MVSplat. The highlighted regions show that OccSplat recovers newly disoccluded content while preserving existing scene structures, whereas DrivingForward leaves holes due to missing Gaussian support and MVSplat exhibits geometric distortions caused by inaccurate depth estimation.
Figure 4: Qualitative comparison of reconstructed 3D Gaussian scenes. Top row: DrivingForward (left) and OccSplat (right). Bottom row: OccSplat with newly added Gaussian primitives highlighted in red (left) and MVSplat (right). The added primitives are distributed along the upper scene boundary, the road foreground, and other regions with incomplete coverage.