Semantic Segmentation Performance (FPS) on ORAD-3D (test)
21.11FPSOT-Drive
Evaluation results
| Method | Links | |
|---|---|---|
| OT-DriveHardware=NVIDIA Tesla T4, Surface normal estimation latency=Included, Inference Optimization=Optimized (PyTorch 2.0 graph compilation and FP16 mixed-precision)2026.01 | 21.11 | |
| OFF-NetHardware=NVIDIA Tesla T4, Surface normal estimation latency=Included, Inference Optimization=Base2026.01 | 4.49 | |
| M2F2-NetHardware=NVIDIA Tesla T4, Surface normal estimation latency=Included, Inference Optimization=Base2026.01 | 4.23 | |
| OT-DriveHardware=NVIDIA Tesla T4, Surface normal estimation latency=Included, Inference Optimization=Base2026.01 | 3.73 | |
| RODHardware=NVIDIA Tesla T4, Surface normal estimation latency=Not Included, Inference Optimization=Base2026.01 | 3.58 |