👉 Try the demo: https://github.com/LM-Kit/lm-kit-net-samples/tree/main/console_net/vision/background-removal/u2net_background_remover

On-Device Background Removal for C# .NET Applications

🎯 Purpose of the Demo

An interactive console app that cuts out the foreground subject and writes a transparent-background PNG. The U2-Net segmentation model loads once at startup, then the menu offers single-image and folder modes.

All processing runs on-device.

👥 Industry Target Audience

E-commerce product photography: bulk-cut catalog photos.
Identity / KYC: clean up document selfies before downstream analysis.
Marketing automation: retouch creatives at scale without per-image cost.
Photo apps: power "remove background" features on user devices.
Compliance: keep customer photos local instead of sending to a cloud API.

🚀 Problem Solved

Cloud background-removal APIs charge per image, require uploading customer photos, and add latency. LM-Kit ships the U2-Net model and a clean API surface: load, call, save. The demo turns it into a prompt-driven app that scales from a single test image to a folder of thousands.

💻 Application Overview

Interactive menu (no command-line arguments) with two modes:

Mode	What it does
File	Cut out a single image. Prompts for input and output path.
Folder	Process every supported image (`.png`, `.jpg`, `.jpeg`, `.bmp`, `.webp`, `.tif`, `.tiff`) in a folder. Output goes to a chosen directory.
Quit	Exit.

The U2-Net model loads once at startup. Each cut-out reports elapsed milliseconds.

✨ Key Features

LM.LoadFromModelID("u2net") for the segmentation model.
BackgroundDetection(model) engine.
RemoveBackground(image) (sync) and RemoveBackgroundAsync(image) (async).
ImageBuffer.LoadAsRGB / SaveAsPng round-trip.

🧠 Model

u2net (bundled segmentation model; ~170 MB, CPU-friendly).

🛠️ Getting Started

📋 Prerequisites

.NET 8.0 or later

▶️ Running the Application

git clone https://github.com/LM-Kit/lm-kit-net-samples
cd lm-kit-net-samples/console_net/vision/background-removal/u2net_background_remover
dotnet run

Pick a mode from the menu and follow the prompts.

Table of Contents