Update README.md
#4
by
reach-vb
HF staff
- opened
README.md
CHANGED
|
@@ -23,6 +23,30 @@ alt="drawing" width="600"/>
|
|
| 23 |
|
| 24 |
<small> Depth Anything overview. Taken from the <a href="https://arxiv.org/abs/2401.10891">original paper</a>.</small>
|
| 25 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
## Download
|
| 27 |
|
| 28 |
Install `huggingface-hub`
|
|
|
|
| 23 |
|
| 24 |
<small> Depth Anything overview. Taken from the <a href="https://arxiv.org/abs/2401.10891">original paper</a>.</small>
|
| 25 |
|
| 26 |
+
## Evaluation - Variants
|
| 27 |
+
|
| 28 |
+
| Variant | Parameters | Size (MB) | Weight precision | Act. precision | abs-rel error | abs-rel reference |
|
| 29 |
+
| ------------------------------------------------------- | ---------: | --------: | ---------------- | -------------- | ------------: | ----------------: |
|
| 30 |
+
| base-original (PyTorch) | 97.5M | 390 | Float32 | Float32 | | |
|
| 31 |
+
| small-original (PyTorch) | 24.8M | 99.2 | Float32 | Float32 | 0.1589 | base-original |
|
| 32 |
+
| [base-float32](depth-anything-base-float32.mlpackage) | 97.5M | 194.6 | Float32 | Float32 | 0.0056 | base-original |
|
| 33 |
+
| [base-float16](depth-anything-base-float16.mlpackage) | 97.5M | 194.6 | Float16 | Float16 | 0.0061 | base-original |
|
| 34 |
+
| [small-float32](depth-anything-small-float32.mlpackage) | 24.8M | 99.0 | Float32 | Float32 | 0.0073 | small-original |
|
| 35 |
+
| [small-float16](depth-anything-small-float16.mlpackage) | 24.8M | 45.8 | Float16 | Float16 | 0.0077 | small-original |
|
| 36 |
+
|
| 37 |
+
## Evaluation - Inference time
|
| 38 |
+
|
| 39 |
+
The following results use the small-float16 variant.
|
| 40 |
+
|
| 41 |
+
| Device | OS | Inference time (ms) | Dominant compute unit |
|
| 42 |
+
| -------------------- | ---- | ------------------: | --------------------- |
|
| 43 |
+
| iPhone 14 | 17.5 | 160.59 | Neural Engine |
|
| 44 |
+
| iPhone 14 Pro Max | 17.5 | 119.33 | Neural Engine |
|
| 45 |
+
| iPhone 15 | 17.0 | 99.42 | Neural Engine |
|
| 46 |
+
| iPhone 15 Pro Max | 17.4 | 116.1 | Neural Engine |
|
| 47 |
+
| MacBook Pro (M1 Max) | 14.5 | 32.20 | GPU |
|
| 48 |
+
|
| 49 |
+
|
| 50 |
## Download
|
| 51 |
|
| 52 |
Install `huggingface-hub`
|