| --- |
| license: apache-2.0 |
| tags: |
| - executorch |
| - xnnpack |
| - pte |
| - on-device |
| - image-segmentation |
| --- |
| # u2net — ExecuTorch |
|
|
| - **Source**: xuebinqin/U-2-Net + Carve/u2net-universal weights |
| - **License**: Apache-2.0 |
| - **Input**: [[1, 3, 320, 320]] — RGB, ImageNet norm, 320x320 |
| - **Output**: saliency mask [1,1,320,320] 0-1 (sigmoid); min-max normalize then resize to the source image |
|
|
| ## 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 | `u2net_xnnpack_fp32.pte` | 176.0 | 1.000000 | 56.5 | |
| | Core ML (fp16, iOS) | `u2net_coreml_all.pte` | 89.0 | 0.999994 | 12.7 | |
|
|
|
|
| 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: 138.5 ms). |
| |
| ### Builds that did not earn a slot |
| |
| - **fp16 is not shipped**: it comes out at 100% of the fp32 file (176.0 MB vs 176.0 MB), so it buys nothing. XNNPACK serializes convolution weights as fp32 no matter what dtype the graph carries, so on a conv-heavy model fp16 saves no disk and only adds cast operations. Reach for int8 here, not fp16. |
| - **int8 is not shipped**: measured in the units that matter for this model — mask IoU at 0.5: median 0.9888 over 10 real images, worst 0.2124. |
| |
| ## 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, 1, 320, 320] | 5.239e-05 | 1.000000 | |
| |
| XNNPACK delegate coverage (fp32): 100.0% (477/477 ops) |
| |
| ## Conversion |
| |
| torch.export -> to_edge_transform_and_lower(partitioner) -> .pte |
| (conversion scripts: [executorch-models](https://github.com/john-rocky/executorch-models)) |
| |
| <!-- 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 --> |
| |
