Upload YuNet-Face-OpenVINO: ONNX, OpenVINO IR, Intel NPU blob, card and manifest

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+npu/yunet_fp16.blob filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,101 @@
+---
+license: bsd-3-clause
+base_model: litert-community/YuNet-Face-LiteRT
+base_model_relation: quantized
+library_name: openvino
+pipeline_tag: object-detection
+tags:
+- openvino
+- onnx
+- intel-npu
+---
+
+# YuNet-Face-OpenVINO
+
+YuNet-Face for Intel NPUs, GPUs and CPUs: yunet_fp16 from [litert-community/YuNet-Face-LiteRT](https://huggingface.co/litert-community/YuNet-Face-LiteRT/tree/f08a6874cf8a1ba9c08ad7a7b9b11231a53b36c9) as ONNX (tf2onnx 1.17.0, opset 17), OpenVINO IR with fp32 weights, and a precompiled Intel NPU blob. On the NPU of a Core Ultra 9 386H the blob runs in 1.56 ms per run, numpy input to numpy output.
+
+Scripts, raw records, and the same table for all 52 graphs of LiteRT-Models: https://github.com/john-rocky/windows-zoo
+
+## Files
+
+| file | bytes | sha256 |
+|---|---|---|
+| `yunet_fp16.onnx` | 433,765 | `898c94029107b6199d2e9c1b3ac2a5603542219d47b2622fa03fb0df4d0ef74d` |
+| `openvino/yunet_fp16.xml` | 595,362 | `24264347235c60446d598052586f6c7c57f4c3860eddb44728e9de47cf49913a` |
+| `openvino/yunet_fp16.bin` | 295,828 | `dfed24938232fb2e22ddc98df13a43bfbd1a8d6333e31f2bd49c961f5bc0e0d6` |
+| `npu/yunet_fp16.blob` | 794,830 | `ffa5169f3d353cd535c1a383062c7e5e59ddff2eb852f0e88a2dcea8cc923472` |
+
+The NPU blob was compiled with OpenVINO 2026.4.1, Intel NPU driver 32.0.100.4841 (OpenVINO reports NPU_DRIVER_VERSION 1004841, NPU_COMPILER_VERSION 524292). If `import_model` refuses a blob on another driver or OpenVINO version, compile the IR instead (the commented line in the snippet).
+
+## How it was measured
+
+MSI Stealth 16 AI+ (Core Ultra 9 386H, RTX 5080 Laptop GPU at a 95 W power limit, Windows 11 build 26200.9711, on AC power), 2026-11-06 to 2026-11-07. On the laptop, each row is one process: load, one run, 5 warmup runs, then the median of 30 timed runs (2 warmup + 10 timed runs when the first run takes more than 3 s), numpy input to numpy output, every runtime at its default settings, one process at a time, each started after the CPU, the RTX and the Intel GPU had gone idle (the thresholds are in METHOD.md of the GitHub repo). Output check: each output against the CPU output in fp32 on the same input (ONNX Runtime's CPU EP); PASS = cosine ≥ 0.99, no NaN or Inf, and top-1 agreement ≥ 0.99 for class scores. The Snapdragon rows are Qualcomm AI Hub profile jobs of the same ONNX files (their own timer).
+
+## Speed and output check
+
+| processor | runtime | file | ms | load s | output check |
+|---|---|---|---|---|---|
+| Intel AI Boost NPU | OpenVINO 2026.4.1 | `npu/` blob (`import_model`) | 1.56 | 0.00 | PASS (cos ≥ 0.9999) |
+| Intel AI Boost NPU | ONNX Runtime 1.31.1 + OpenVINO EP 1.8.1 | `.onnx`, compiled when the session loads | 1.76 | 1.06 | PASS (cos ≥ 0.9999) |
+| Intel Graphics (iGPU) | ONNX Runtime 1.31.1 + OpenVINO EP 1.8.1 | `.onnx` | 1.58 | 0.42 | PASS (cos ≥ 0.9999) |
+| RTX 5080 Laptop (95 W limit) | ONNX Runtime 1.31.1, CUDA EP | fp16 copy of the `.onnx` (not in this repo) | 6.19 | 1.20 | PASS (cos ≥ 0.9999) |
+| RTX 5080 Laptop (95 W limit) | ONNX Runtime 1.31.1, CUDA EP | `.onnx` | 2.83 | 1.22 | PASS (cos ≥ 0.9999) |
+| Core Ultra 9 386H CPU | ONNX Runtime 1.31.1, CPU EP | `.onnx` | 3.10 | 0.12 | reference |
+| Snapdragon X2 Elite NPU | Qualcomm AI Hub, precompiled QNN ONNX | `.onnx` compiled by AI Hub | 1.32 | 0.46 | not compared (AI Hub profile job, 2026-10-30) |
+| Snapdragon X2 Elite NPU | Qualcomm AI Hub, ONNX Runtime QNN EP | `.onnx`, compiled when it loads | 2.04 (2 layers on the CPU) | 1.11 | not compared (AI Hub profile job, 2026-10-30) |
+| Galaxy S26 NPU | Qualcomm AI Hub, precompiled QNN ONNX | `.onnx` compiled by AI Hub | 1.29 | 0.14 | not compared (AI Hub profile job, 2026-10-30) |
+
+The ONNX file against the source `.tflite` (LiteRT interpreter on the CPU, the same input): PASS, cosine ≥ 0.9999 for every output.
+
+## Run it on the Intel NPU
+
+Download this repo, then run the snippets from its folder:
+
+```sh
+pip install huggingface_hub
+hf download mlboydaisuke/YuNet-Face-OpenVINO --local-dir YuNet-Face-OpenVINO
+cd YuNet-Face-OpenVINO
+```
+
+OpenVINO (`pip install openvino==2026.4.1`):
+
+```python
+import numpy as np
+import openvino as ov
+
+core = ov.Core()
+if "NPU" not in core.available_devices:
+    raise SystemExit(f"no Intel NPU visible to OpenVINO: {core.available_devices}")
+with open("npu/yunet_fp16.blob", "rb") as f:
+    compiled = core.import_model(f.read(), "NPU")  # precompiled: no compile at load
+# or compile the IR on this machine: compiled = core.compile_model("openvino/yunet_fp16.xml", "NPU")
+# replace the zeros with your preprocessed input (float32, the shapes printed by compiled.inputs)
+feeds = {i.get_any_name(): np.zeros(list(i.shape), np.float32) for i in compiled.inputs}
+outputs = compiled(feeds)
+print({o.get_any_name(): outputs[o].shape for o in compiled.outputs})
+```
+
+ONNX Runtime with the OpenVINO EP (`pip install onnxruntime==1.31.1 onnxruntime-ep-openvino==1.8.1`):
+
+```python
+import numpy as np
+import onnxruntime as ort
+import onnxruntime_ep_openvino as ov_ep
+
+ort.register_execution_provider_library(ov_ep.get_ep_name(), ov_ep.get_library_path())
+npu = [d for d in ort.get_ep_devices()
+       if d.ep_name == ov_ep.get_ep_name() and d.ep_metadata.get("ov_device") == "NPU"]
+if not npu:
+    raise SystemExit("no Intel NPU visible to the OpenVINO EP")
+so = ort.SessionOptions()
+so.add_provider_for_devices(npu, {})
+sess = ort.InferenceSession("yunet_fp16.onnx", sess_options=so)  # compiles for the NPU here
+# replace the zeros with your preprocessed input (float32)
+feeds = {i.name: np.zeros(i.shape, np.float32) for i in sess.get_inputs()}
+outputs = sess.run(None, feeds)
+print([o.shape for o in outputs])
+```
+
+## Licence
+
+BSD-3-Clause, the licence of [litert-community/YuNet-Face-LiteRT](https://huggingface.co/litert-community/YuNet-Face-LiteRT). The source card names the upstream model as `ShiqiYu/libfacedetection` (not a Hugging Face repo).

manifest.json

@@ -0,0 +1,39 @@
+{
+ "source_repo": "litert-community/YuNet-Face-LiteRT",
+ "source_revision": "f08a6874cf8a1ba9c08ad7a7b9b11231a53b36c9",
+ "graphs": {
+  "yunet_fp16": {
+   "from": "yunet_fp16",
+   "ir": {
+    "openvino": "2026.4.1-22982-e213a147257-releases/2026/4",
+    "compress_to_fp16": false
+   },
+   "blob_compiled_with": {
+    "openvino": "2026.4.1-22982-e213a147257-releases/2026/4",
+    "FULL_DEVICE_NAME": "Intel(R) AI Boost",
+    "NPU_DRIVER_VERSION": "1004841",
+    "NPU_COMPILER_VERSION": "524292",
+    "NPU_COMPILER_TYPE": "CompilerType.PREFER_PLUGIN",
+    "INFERENCE_PRECISION_HINT": "<Type: 'float16'>"
+   },
+   "files": {
+    "yunet_fp16.onnx": {
+     "bytes": 433765,
+     "sha256": "898c94029107b6199d2e9c1b3ac2a5603542219d47b2622fa03fb0df4d0ef74d"
+    },
+    "openvino/yunet_fp16.xml": {
+     "bytes": 595362,
+     "sha256": "24264347235c60446d598052586f6c7c57f4c3860eddb44728e9de47cf49913a"
+    },
+    "openvino/yunet_fp16.bin": {
+     "bytes": 295828,
+     "sha256": "dfed24938232fb2e22ddc98df13a43bfbd1a8d6333e31f2bd49c961f5bc0e0d6"
+    },
+    "npu/yunet_fp16.blob": {
+     "bytes": 794830,
+     "sha256": "ffa5169f3d353cd535c1a383062c7e5e59ddff2eb852f0e88a2dcea8cc923472"
+    }
+   }
+  }
+ }
+}

npu/yunet_fp16.blob

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ffa5169f3d353cd535c1a383062c7e5e59ddff2eb852f0e88a2dcea8cc923472
+size 794830

openvino/yunet_fp16.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dfed24938232fb2e22ddc98df13a43bfbd1a8d6333e31f2bd49c961f5bc0e0d6
+size 295828

yunet_fp16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:898c94029107b6199d2e9c1b3ac2a5603542219d47b2622fa03fb0df4d0ef74d
+size 433765