AI Paper Atlas

Research map / Detection and segmentation / CNN features

SNE-RoadSeg: Incorporating Surface Normal Information into Semantic Segmentation for Accurate Freespace Detection

Rui Fan, Hengli Wang, Peide Cai, Ming Liu

ECCV 2020 · Detection and segmentation › CNN features · approach: Lane detection · 201 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add surface normal information from SNE to depth input (ablation)RoadSeg-152 depth only · R2D road datasetIoU (%): 92.4 -> 94.5Improved
add SNE to four-channel RGB-D input (ablation)RoadSeg-152 single encoder · R2D road datasetIoU (%): 89.0 -> 93.0Improved
use two-encoder fusion instead of four-channel input (ablation)RoadSeg-152 with SNE · R2D road datasetIoU (%): 93.0 -> 96.7Improved
add depth encoder to RGB-only model (ablation)RoadSeg-152 · R2D road datasetIoU (%): 86.6 -> 96.7Improved
use ResNet-152 instead of ResNet-18 backbone (ablation)RoadSeg · R2D road datasetIoU (%): 85.9 -> 96.7Improved
use ResNet-152 instead of ResNet-101 backbone (ablation)RoadSeg · R2D road datasetF-score (%): 97.6 -> 98.3Improved
add dense skip connections in decoder instead of none (ablation)RoadSeg-152 · R2D road datasetIoU (%): 95.2 -> 96.7Improved
use dense skip connections instead of U-Net-style sparse ones (ablation)RoadSeg-152 · R2D road datasetIoU (%): 95.9 -> 96.7Improved
embed SNE into single-modal CNNsten state-of-the-art CNNs · SYNTHIA and R2DIoU (%): approximately 2-12%Improved
embed SNE into data-fusion CNNsdata-fusion CNNs · SYNTHIA and R2DIoU (%): about 1-7%Improved
use RoadSeg instead of other CNNs, all with SNERoadSeg with SNE · SYNTHIA and R2DIoU (%): approximately 1.4-14.7%Improved
compare SNE-RoadSeg with RBANet on KITTISNE-RoadSeg · KITTI road benchmarkMaxF (%): 96.30 -> 96.75Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy