FastDeploy/examples/vision/classification/paddleclas/web

English | 简体中文

MobileNet Front-end Deployment Example

This document introduces the deployment of PaddleClas's mobilenet models for image classification to run in the browser, and the js interface in the @paddle-js-models/mobilenet npm package.

Front-end Deployment of Image Classification Model

To use the web demo of image classification models, refer to Reference Document

MobileNet js Interface

import * as mobilenet from "@paddle-js-models/mobilenet";
# mobilenet model loading and initialization
await mobilenet.load()
# mobilenet model performs the prediction and obtains the classification result
const res = await mobilenet.classify(img);
console.log(res);

load() function parameter

• Config(dict): The configuration parameter for the image classification model. Default {Path: 'https://paddlejs.bj.bcebos.com/models/fuse/mobilenet/mobileNetV2_fuse_activation/model.json', fill: '#fff', mean: [0.485, 0.456, 0.406],std: [0.229, 0.224, 0.225]}; Among them, modelPath is the path of the js model, fill is the padding value in the image pre-processing, and mean/std are the mean and standard deviation in the pre-processing

classify() function parameter

• img(HTMLImageElement): Enter an image parameter in HTMLImageElement.

Other Documents

• PaddleClas model python deployment
• PaddleClas model C++ deployment