MACHINE LEARNING. HUMAN INTENT.

A more
considered
web.

Exploring how machine learning can help put a little more control between you and the content you see.

XDynamic Lab Browser technology & machine learning

a more considered web
Room to choose.Content, with a layer of control.
ABSTRACT STUDY / 001
Blur preview on

Interactive concept illustration
Sample artwork. No live detection.

THE EXPLORATIONBrowser extensionsMachine learningContent filtering↓

Technology with
a clear intention.

The web is full of things we choose to see—and things we don’t. XDynamic Lab introduces a project exploring that boundary.

Our focus here is XDynamic Extension: a browser extension project for detecting and blurring sensitive web content with machine learning. Explore its implementation, understand its architecture, and follow the code back to its origins.

Meet the technology

A LAYER BETWEEN YOU AND THE UNEXPECTED

XDynamic Extension.

Explore the repository

A browser extension project focused on identifying sensitive (NSFW) web content and applying a blur to flagged images.

01

Detect the sensitive.

Machine learning inference evaluates images for sensitive content as part of the project’s detection flow.

02

Add a little distance.

The extension’s content script applies a visual blur to images identified for filtering.

03

Explore the implementation.

Follow the browser extension, backend API, and admin dashboard in the public project repository.

PROJECT NOTE

See the repository for setup and current capabilities. The extension uses a backend for ML inference; this page is an introduction to the project.

Connected pieces.
Considered design.

The repository separates the browser experience from the API and model inference. Each part has a distinct role.

Read the architecture
01

Browser experience

React · TypeScript · Manifest V3

A browser extension targeting Chrome and Firefox.
02

Application layer

Python · FastAPI

An API connecting the extension to backend services.
03

Machine learning

PyTorch · MobileNetV2

Image classification in the ML inference service.

04 / BUILT ON SHARED WORK

The code tells
the whole story.

Look through the source, explore the documentation, and understand how the pieces come together.

View project on GitHub
REPOSITORY / PUBLIC

anhht9824 /

xdynamic-extension

WITH CREDIT TO THE ORIGINAL PROJECT

This repository is a fork of nekloyh/xdynamic-extension . Credit for the upstream work belongs to its original authors and contributors.

Visit upstream

05 / START A CONVERSATION

Good ideas begin
with a hello.

Questions about the project or an idea to share?
Get in touch with XDynamic Lab.

founder@xdynamiclab.xyz