I Open-Sourced a Browser-Based AI Background Remover — Here's the Full Architecture
The open-source background remover uses @imgly/background-removal, which bundles an ONNX segmentation model with ONNX Runtime Web (WebAssembly backend). The core logic consists of two files with zero dependencies beyond a CDN import from jsDelivr. The first call downloads approximately 40MB of model weights, which the browser caches automatically for instant subsequent loads. The full pipeline runs entirely client-side through five stages: upload, ONNX model load, WebAssembly inference, mask generation, and canvas compositing. The tool ensures user images never leave the device, addressing privacy concerns of server-based alternatives.
// why it matters
Client-side AI processing eliminates server costs and privacy risks for image editing tools.