Home News Latest Papers Tutorials Datasets Wiki SOTA LLM Models GPU Leaderboard Events

English

3D Lane Detection On Openlane

Metrics

Curve

Extreme Weather

F1 (all)

Intersection

Merge u0026 Split

Night

Up u0026 Down

Results

Performance results of various models on this benchmark

Model Name	Curve	Extreme Weather	F1 (all)	Intersection	Merge u0026 Split	Night	Up u0026 Down	Paper Title	Repository
Anchor3DLane† (iterative regression)	57.2	52.5	53.7	45.4	51.2	47.8	46.7	Anchor3DLane: Learning to Regress 3D Anchors for Monocular 3D Lane Detection	-
PVALane (Swin-B)	67.7	64.0	63.4	53.6	60.8	58.6	56.1	PVALane: Prior-Guided 3D Lane Detection with View-Agnostic Feature Alignment	-
CurveFormer	56.6	49.7	50.5	42.9	45.4	49.1	45.2	CurveFormer: 3D Lane Detection by Curve Propagation with Curve Queries and Attention	-
PersFormer (version 1.2)	58.4	51.8	52.9	42.1	50.9	47.4	47.5	PersFormer: 3D Lane Detection via Perspective Transformer and the OpenLane Benchmark	-
LATR	68.2	57.1	61.9	52.3	61.5	55.4	55.2	LATR: 3D Lane Detection from Monocular Images with Transformer	-
PETRv2-V∗ (VoVNetV2 with 400 anchor points)	-	-	61.2	-	-	-	-	PETRv2: A Unified Framework for 3D Perception from Multi-Camera Images	-
PETRv2-E (EfficientNet)	-	-	51.9	-	-	-	-	PETRv2: A Unified Framework for 3D Perception from Multi-Camera Images	-
RFTR (ResNet-50)	-	-	61.8	-	-	-	-	RepVF: A Unified Vector Fields Representation for Multi-task 3D Perception	-
3D-LaneNet	46.5	47.5	44.1	32.1	41.7	41.5	40.8	3D-LaneNet: End-to-End 3D Multiple Lane Detection	-
PVALane (ResNet-50)	67.3	62.0	62.7	53.4	60.0	57.2	54.1	PVALane: Prior-Guided 3D Lane Detection with View-Agnostic Feature Alignment	-
Anchor3DLane (ResNet-18)	56.2	51.9	53.1	44.2	50.5	47.2	45.5	Anchor3DLane: Learning to Regress 3D Anchors for Monocular 3D Lane Detection	-
PETRv2-V (VoVNetV2)	-	-	57.8	-	-	-	-	PETRv2: A Unified Framework for 3D Perception from Multi-Camera Images	-
LaneCPP	64.4	56.7	60.3	52.0	58.7	54.9	53.6	LaneCPP: Continuous 3D Lane Detection using Physical Priors	-
M^2-3DLaneNet (Camera + Lidar)	60.7	56.2	55.5	43.8	51.4	51.6	53.4	M$^2$-3DLaneNet: Exploring Multi-Modal 3D Lane Detection	-
PVALane (ResNet-18)	65.7	59.5	61.2	52.2	58.7	56.5	52.6	PVALane: Prior-Guided 3D Lane Detection with View-Agnostic Feature Alignment	-
BEV-LaneDet	63.1	53.4	58.4	50.3	53.7	53.4	48.7	BEV-LaneDet: a Simple and Effective 3D Lane Detection Baseline	-
PersFormer (version 1.1)	58.7	54.0	50.5	41.6	53.1	50.0	45.6	PersFormer: 3D Lane Detection via Perspective Transformer and the OpenLane Benchmark	-
Gen-LaneNet	33.5	28.1	32.3	21.4	31.0	18.7	25.4	Gen-LaneNet: A Generalized and Scalable Approach for 3D Lane Detection	-
Anchor3DLane-T† (multi-frame + iterative regression)	58.0	52.7	54.3	45.8	51.7	48.7	47.2	Anchor3DLane: Learning to Regress 3D Anchors for Monocular 3D Lane Detection	-

0 of 19 row(s) selected.

3D Lane Detection On Openlane | SOTA | HyperAI