Upload folder using huggingface_hub

README.md

@@ -1,3 +1,111 @@
 ---
 license: mit
+library_name: onnx
+tags:
+  - image-denoising
+  - onnx
+  - drunet
+  - raw-photography
 ---
+
+# DRUNet (colour) — ONNX export
+
+An ONNX export of **DRUNet**, the plug-and-play denoiser from Kai Zhang's
+[KAIR](https://github.com/cszn/KAIR) toolbox. Nothing here is retrained or modified: this is the
+original `drunet_color.pth` weights, exported to ONNX so they can be run from ONNX Runtime.
+
+Published for use in [PxlMonk](https://pxlmonk.com), a desktop raw photo editor, whose build fetches
+this file at build time rather than committing a 130 MB binary.
+
+## Provenance
+
+| | |
+|---|---|
+| Architecture | `UNetRes` from [cszn/KAIR](https://github.com/cszn/KAIR) — MIT |
+| Weights | `drunet_color.pth`, from the KAIR v1.0 release — unmodified |
+| Export | `torch.onnx.export`, opset 17, dynamic batch/height/width |
+| Paper | Zhang et al., *Plug-and-Play Image Restoration with Deep Denoiser Prior*, TPAMI 2021 |
+
+The weights were trained by the original author and are redistributed unchanged, in a different file
+format. No training data is redistributed here.
+
+### A note on the training corpora
+
+DRUNet was trained on `train400`, **DIV2K**, **Flickr2K** and the Waterloo Exploration Database. Some of
+those carry their own terms — DIV2K states:
+
+> This dataset is made available for academic research purpose only. All the images are collected from
+> the Internet, and the copyright belongs to the original owners.
+
+Whether such dataset terms reach the resulting model weights is unsettled, and this repository takes no
+position on it. What it does do is state the provenance plainly, so anyone using these weights can make
+that call with the facts in front of them rather than discovering them later. The MIT tag above reflects
+the licence under which the original author published the weights.
+
+## Interface
+
+Input `input`: `float32[batch, 4, height, width]`, NCHW.
+
+- Channels 0–2: RGB, **sRGB display-referred**, range 0…1. The model was trained on display images —
+  feeding scene-linear data gives poor results.
+- Channel 3: the **noise level map**, a constant `sigma / 255` across the plane. This is DRUNet's
+  strength control: sigma is in 0…255 units, so `sigma = 20` means a plane filled with `20/255`.
+
+Output `output`: `float32[batch, 3, height, width]`, same convention, denoised.
+
+Height and width must be divisible by 8 (three downsampling stages). Pad or tile accordingly.
+
+## Choosing sigma
+
+There is no single right value — it is the strength dial. Measured on a 24 MP frame shot at ISO 16000,
+using mean deviation from the local 3×3 mean as a noise figure (lower is cleaner):
+
+| sigma | noise figure | |
+|---|---|---|
+| — (untouched) | 7.41 | |
+| 10 | 1.67 | visible noise remains |
+| **20** | **0.23** | clean, fine detail intact |
+| 35 | 0.17 | over-smoothed — eyelashes go waxy |
+
+## Reproducing this export
+
+The weights are stored **without bias terms**, so the network must be constructed with `bias=False`.
+Loading with `strict=False` silently leaves 64 randomly-initialised bias tensors in place and produces
+a model that runs but denoises badly — always load strictly.
+
+```python
+from models.network_unet import UNetRes   # from cszn/KAIR
+import torch
+
+net = UNetRes(in_nc=4, out_nc=3, nc=[64, 128, 256, 512], nb=4, act_mode="R",
+              downsample_mode="strideconv", upsample_mode="convtranspose", bias=False)
+net.load_state_dict(torch.load("drunet_color.pth", map_location="cpu", weights_only=True), strict=True)
+net.eval()
+
+torch.onnx.export(
+    net, torch.randn(1, 4, 256, 256), "drunet_color.onnx",
+    input_names=["input"], output_names=["output"],
+    dynamic_axes={"input": {0: "b", 2: "h", 3: "w"}, "output": {0: "b", 2: "h", 3: "w"}},
+    opset_version=17,
+)
+```
+
+## Execution providers
+
+Measured on Apple Silicon (10 cores), 256×256 tile, best of three after a warm-up:
+
+| | |
+|---|---|
+| CPU | 560 ms |
+| CoreML | **117 ms** |
+| WebGPU | 154 ms |
+
+DRUNet is a plain CNN, so it maps cleanly onto accelerators — unlike transformer-based denoisers, where
+the graph fragments into hundreds of partitions and the copying between them costs more than the
+compute saves.
+
+## Licence
+
+MIT, matching KAIR. Include the original copyright notice when redistributing:
+
+> Copyright (c) 2019 Kai Zhang

drunet_color.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ae3ab5eb15daac2ee79be984d584b908ce7f0f60b27be87d005f728c2aa0087
+size 130589232

export_drunet_onnx.py

@@ -0,0 +1,28 @@
+# DRUNet (KAIR, MIT) nach ONNX. Die Gewichte sind cszns Original; Architektur aus demselben Repo,
+# damit die Schlüssel ohne Umbenennen passen.
+import sys, torch
+sys.path.insert(0, "/private/tmp/claude-501/-Users-ckret-wkspaces-pxlmonk/35a64bed-d9bb-4a13-9d30-2c08f92e2fd2/scratchpad/kair")
+from models.network_unet import UNetRes
+
+SP = "/private/tmp/claude-501/-Users-ckret-wkspaces-pxlmonk/35a64bed-d9bb-4a13-9d30-2c08f92e2fd2/scratchpad"
+net = UNetRes(in_nc=4, out_nc=3, nc=[64,128,256,512], nb=4, act_mode='R',
+              downsample_mode="strideconv", upsample_mode="convtranspose", bias=False)
+sd = torch.load(f"{SP}/kair/drunet_color.pth", map_location="cpu", weights_only=True)
+net.load_state_dict(sd, strict=True)   # strict: jede Abweichung ist ein Fehler, kein Zufallsgewicht
+missing, unexpected = [], []
+print("fehlend:", len(missing), " unerwartet:", len(unexpected))
+if missing or unexpected:
+    print("  ", missing[:3], unexpected[:3])
+net.eval()
+
+# 4. Kanal ist die Rauschpegelkarte: konstant sigma/255 über das ganze Bild.
+dummy = torch.randn(1, 4, 256, 256)
+out = f"{SP}/models/drunet_color.onnx"
+torch.onnx.export(
+    net, dummy, out,
+    input_names=["input"], output_names=["output"],
+    dynamic_axes={"input": {0: "b", 2: "h", 3: "w"}, "output": {0: "b", 2: "h", 3: "w"}},
+    opset_version=17,
+)
+import os
+print(f"geschrieben: {out}  ({os.path.getsize(out)/1e6:.1f} MB)")