V2X-Real
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
V2X-Real EGO+AUX1 1.0 (seen-pair)
52.5mAP@0.5
80
V2X-Real EGO+AUX2 (unseen evaluation)
49.9mAP@0.5
80
V2X-Real EGO+AUX2 zero-shot 128-beam LiDAR
52mAP@0.5
52
V2X-Real (test)
75.88AP @ IoU=0.3
18
V2X-Real (test)
91.3Vehicle AP@0.3
18
V2X-Real
54.04mAP@0.3
17
V2X-Real (EGO+AUX1)
54.3E2E mAP@0.5
16
V2X-Real EGO+AUX2 unseen adaptation 1.0 (test)
47.9E2E mAP@0.5
16
V2X-Real Gaussian Pose Noise σ = 0.4 (test)
84.27AP@0.3 (Vehicle)
16
V2X-Real Dynamic-only
25.93PSNR
9
V2X-Real Full image
25.89PSNR
9
V2X-Real Infrastructure-Centric
86.7Car AP@0.3
9
V2X-Real Vehicle-Centric
91.5Car AP@0.3
9
V2X-Real Ideal Setting (No Noise) (test)
89.63AP@0.3 (Vehicle)
8
V2X-Real Ideal Setting
89.63AP@0.3 (Vehicle)
8
V2X-Real Gaussian Pose Noise σ = 0.4
84.27AP@0.3 (vehicle)
8
V2X-Real Ideal Setting (test)
89.63AP@0.3 (Vehicle)
8
V2X-Real
53.8mAP@30
6
V2X-Real
46.02mAP30 (Overall)
6
V2X-Real 25
40.1AP@0.3
6
V2X-Real Jointly Observed
47.85mAP@0.3
5
V2X-Real Overall
38.6mAP@0.3
5
V2X-Real
83.6CLIP Similarity
5
V2X-Real (val)
11.48mAP@0.3
5
V2X-Real (test)
21.47Full Image PSNR
4