WebNN Samples

See samples built with WebNN API. Usually the best and recommended way to leverage WebNN capabilities and performance for your web applications is to use TensorFlow.js or other JavaScript library that integrates WebNN API directly. You can also build web applications powered by WebNN from scratch without using ML JavaScript libraries.

WebNN Samples

continuously updated to follow the latest WebNN spec

WebNN Code Editor

Evaluate, review and modify WebNN sample codes interactively in web browser.

Image Classfication

Classifying the major object in the image into a set of pre-defined classes.

Handwritten Digits Classification

Use LeNet to classify handwritten digits from the clasic MNIST example data set.

Fast Style Transfer

Applying the STARRY NIGHT and other painting styles of Van Gogh to your image.

Noise Suppression (NSNet2)

A NSNet2 baseline implementation of deep learning-based noise suppression.

Noise Suppression (RNNoise)

An RNNoise baseline implementation of deep learning-based noise suppression.

Object Detection

Detecting instances of semantic objects of a certain class in digital images and videos.

Semantic Segmentation

Partitioning image into semantically meaningful parts to classify each part into pre-determined class.

Facial Landmark Detection

Detecting facial landmarks like eyes, nose, mouth, etc., which can be used for web-based try-on simulator of online store.

Face Recognition

Detecting faces of participants using object detection and checking whether each face was present or not.

Legacy WebNN Examples

the following WebNN examples are not up to date but could inspire you for your ideas