Xenova HF staff commited on
Commit
7069a32
1 Parent(s): 1630fe9

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +52 -0
README.md ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers.js
3
+ tags:
4
+ - image-matting
5
+ - portrait-matting
6
+ ---
7
+
8
+ https://github.com/ZHKKKe/MODNet with ONNX weights to be compatible with Transformers.js.
9
+
10
+ ## Usage (Transformers.js)
11
+
12
+ If you haven't already, you can install the [Transformers.js](https://huggingface.co/docs/transformers.js) JavaScript library from [NPM](https://www.npmjs.com/package/@xenova/transformers) using:
13
+ ```bash
14
+ npm i @xenova/transformers
15
+ ```
16
+
17
+ You can then use the model for portrait matting, as follows:
18
+
19
+ ```js
20
+ import { AutoProcessor, RawImage, AutoModel } from '@xenova/transformers';
21
+
22
+ // Load model and processor
23
+ const model = await AutoModel.from_pretrained('Xenova/modnet-onnx', { quantized: false });
24
+ const processor = await AutoProcessor.from_pretrained('Xenova/modnet-onnx');
25
+
26
+ // Load image from URL
27
+ const url = 'https://images.pexels.com/photos/5965592/pexels-photo-5965592.jpeg?auto=compress&cs=tinysrgb&w=1024';
28
+ const image = await RawImage.fromURL(url);
29
+
30
+ // Pre-process image
31
+ const { pixel_values } = await processor(image);
32
+
33
+ // Predict alpha matte
34
+ const { output } = await model({ input: pixel_values });
35
+
36
+ // Save output mask
37
+ const mask = await RawImage.fromTensor(output[0].mul(255).to('uint8')).resize(image.width, image.height);
38
+ mask.save('mask.png');
39
+ ```
40
+
41
+ | Input image | Output mask |
42
+ |--------|--------|
43
+ | ![image/png](https://cdn-uploads.huggingface.co/production/uploads/61b253b7ac5ecaae3d1efe0c/mhmDJgp5GgnbvQnUc2SVI.png) | ![image/png](https://cdn-uploads.huggingface.co/production/uploads/61b253b7ac5ecaae3d1efe0c/H1VBX6dS-xTpg14cl1Zxx.png) |
44
+
45
+ ---
46
+
47
+ Note: Having a separate repo for ONNX weights is intended to be a temporary solution until WebML gains more traction. If you would like to make your models web-ready, we recommend converting to ONNX using [🤗 Optimum](https://huggingface.co/docs/optimum/index) and structuring your repo like this one (with ONNX weights located in a subfolder named `onnx`).
48
+
49
+
50
+
51
+
52
+ For more information, see the original [repo](https://github.com/ZHKKKe/MODNet) and example [colab](https://colab.research.google.com/drive/1P3cWtg8fnmu9karZHYDAtmm1vj1rgA-f?usp=sharing).