Depth estimation: Monocular, relative
Relative depth from one image.
No checkpoint here has both a measured speed and a measured accuracy yet, so there is nothing to plot.
4 measured checkpoints are not yet scored for accuracy.
All monocular, relative models
Each model runs at its official default input, shown under its latency. A smaller input is less work, so compare latencies across input sizes with care.
| Model | Accuracy (paper under it) | 4090 latency (input under it) | VRAM peak | License |
|---|---|---|---|---|
| Depth Anything 3 Mono1 size: large | Not yet scored | 27.3 ms504x336 | 2.1 GiB | Commercial OK |
| Depth Anything V23 sizes: small, base, large | Not yet scored | 17.1 ms770x518 | 858 MiB | Commercial OK smallNon-commercial base, large |