Welcome to the Playground

Run computer vision models directly in your browser, and explore real-time detection, classification, and visual understanding without any setup.

Features

Trainable Vision Models

Create and train custom computer vision models using your own data. (COMING SOON)

Instant Image Analysis

Use ready-made analytical models to inspect images, extract visual properties, and compute insights.

Real-Time AI Sessions

Run both trainable and analytical models in live sessions with low-latency inference powered by WebSockets.

How it works

[1]

Enable Camera Access

Grant permission for your camera. All processing runs entirely in your browser—no video data is ever uploaded or stored.

[2]

Select a Vision Model

Choose from a curated set of optimized computer vision models designed for detection, classification, and tracking.

[3]

Initialize the Model

The selected model is loaded and optimized using browser-native acceleration such as WebGL or WebGPU.

[4]

Real-Time Inference

Live camera frames are analyzed in real time, with predictions generated instantly and visualized on-screen.

[5]

Visualize & Interact

View detection overlays, confidence scores, and results instantly—adjust models or inputs without restarting.

ML-Hub

Built with NextJS and FastAPI © 2026 ML-Hub