File size: 2,152 Bytes
56573a6 c0b30ea 58233b5 56573a6 c0b30ea 58233b5 c0b30ea 27bd167 c0b30ea |
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 |
---
library_name: transformers.js
pipeline_tag: zero-shot-image-classification
license: other
tags:
- mobileclip
- image-feature-extraction
- feature-extraction
---
https://github.com/apple/ml-mobileclip with ONNX weights to be compatible with Transformers.js.
## Usage (Transformers.js)
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:
```bash
npm i @xenova/transformers
```
**Example:** Perform zero-shot image classification.
```js
import {
AutoTokenizer,
CLIPTextModelWithProjection,
AutoProcessor,
CLIPVisionModelWithProjection,
RawImage,
dot,
softmax,
} from '@xenova/transformers';
const model_id = 'Xenova/mobileclip_b';
// Load tokenizer and text model
const tokenizer = await AutoTokenizer.from_pretrained(model_id);
const text_model = await CLIPTextModelWithProjection.from_pretrained(model_id);
// Load processor and vision model
const processor = await AutoProcessor.from_pretrained(model_id);
const vision_model = await CLIPVisionModelWithProjection.from_pretrained(model_id, {
quantized: false, // NOTE: vision model is sensitive to quantization.
});
// Run tokenization
const texts = ['cats', 'dogs', 'birds'];
const text_inputs = tokenizer(texts, { padding: 'max_length', truncation: true });
// Compute text embeddings
const { text_embeds } = await text_model(text_inputs);
const normalized_text_embeds = text_embeds.normalize().tolist();
// Read image and run processor
const url = 'https://huggingface.co/datasets/Xenova/transformers.js-docs/resolve/main/cats.jpg';
const image = await RawImage.read(url);
const image_inputs = await processor(image);
// Compute vision embeddings
const { image_embeds } = await vision_model(image_inputs);
const normalized_image_embeds = image_embeds.normalize().tolist();
// Compute probabilities
const probabilities = normalized_image_embeds.map(
x => softmax(normalized_text_embeds.map(y => 100 * dot(x, y)))
);
console.log(probabilities); // [[ 0.999993040175817, 0.000006828091823929405, 1.3173235896278122e-7 ]]
```
|