File size: 3,371 Bytes
d14beed 61e938a d14beed 61e938a d14beed 61e938a 6ccbd2c d14beed 61e938a d14beed 61e938a d14beed bf8fc86 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 | ---
license: apache-2.0
tags:
- executorch
- xnnpack
- pte
- on-device
- keypoint-detection
- pose-estimation
---
# rtmpose_m_face — ExecuTorch
- **Source**: open-mmlab/mmpose RTMPose (rtmpose-m_simcc-face6_pt-in1k_120e-256x256-72a37400)
- **License**: Apache-2.0
- **Input**: [[1, 3, 256, 256]] — RGB, ImageNet norm, 256x256 crop around a face (detect first, then crop and resize to this aspect)
- **Output**: SimCC pair: x [1,106,512] and y [1,106,512] — a 1-D distribution per keypoint per axis. Decode: keypoint k sits at (argmax(x[k]) / 2, argmax(y[k]) / 2) in crop pixels; the max value doubles as the confidence.
## Variants
All variants take and return fp32 tensors — swap the `.pte` file, keep your app code.
| build | file | size (MB) | parity vs fp32 eager (worst corr) | Mac median (ms)* |
|-----------|------|-----------|------------------------------------|------------------|
| fp32 | `rtmpose_m_face_xnnpack_fp32.pte` | 67.9 | 1.000000 | 10.4 |
| Core ML (fp16, iOS) | `rtmpose_m_face_coreml_all.pte` | 34.4 | 0.999897 | 3.3 |
The Core ML build is the same graph lowered to Apple's Neural Engine instead of
XNNPACK, which is CPU-only. Measured on an iPhone 17 Pro across seven models, it
runs **3.5x to 13.9x faster (median 12x)** at roughly half the file size — for
example Depth-Anything-V2-Small at 500.8 ms against 42.7 ms, and MODNet at 81.7 ms
against 5.9 ms. It computes in fp16 and is iOS-only; the XNNPACK files stay the
portable option and are what runs on Android.
\*Mac arm64, single process, median of 10 — a reference point for relative cost
only, not a device number (torch eager fp32 on the same machine: 94.4 ms).
### Checked in the task's own units
Correlation is a first filter. These are the numbers that decide:
- **Core ML (fp16, iOS)** — measured in the units that matter for this model — fraction of keypoints landing within 4 px of fp32: median 0.9811 over 10 real images, worst 0.9151.
## Verification (executorch 1.4.0, torch 2.13.0)
Parity is measured against the fp32 eager model on real image input; `corr` is
the correlation over all elements of each output tensor.
| output | shape | max_abs_diff | corr |
|--------|-------|--------------|------|
| 0 | [1, 106, 512] | 1.147e-05 | 1.000000 |
| 1 | [1, 106, 512] | 9.179e-06 | 1.000000 |
XNNPACK delegate coverage (fp32): 93.9% (306/326 ops); ops left on the portable kernels: `dim_order_ops._to_dim_order_copy.default` x8, `aten.unsqueeze_copy.default` x3, `aten.split_with_sizes_copy.default` x3, `aten.sum.dim_IntList` x2, `aten.pow.Tensor_Scalar` x2, `aten.squeeze_copy.dims` x2
## Conversion
torch.export -> to_edge_transform_and_lower(partitioner) -> .pte
(conversion scripts: [executorch-models](https://github.com/john-rocky/executorch-models))
**Notes**: Top-down: crop one face first. In practice a portrait framed on the head works directly, which is how the card check exercises it.
<!-- funnel:v1 -->
---
**More models in this format:** [ExecuTorch Model Zoo](https://huggingface.co/collections/mlboydaisuke/executorch-model-zoo-6a7ff328390b63075ffeae5e) — 31 models, each with the recipe that produced it.
**Want a different model on-device?** [Open a request](https://github.com/john-rocky/on-device-requests) — free, open weights only; the export and its measured numbers get published publicly.
<!-- /funnel:v1 -->
|