背景の切り抜き
商品写真や人物写真、モックアップの背景をブラウザの中で切り抜きます。AI セグメンテーションモデルはローカルで動き、画像はサーバーに送られません。
Drop an image here
or click to select a file, or paste
Processing happens entirely in your browser. On first use, a ~180MB processing engine is downloaded and cached.
Background Remover runs an image-segmentation model directly in your browser to produce a transparent PNG of the main subject. The first run downloads the model weights into the browser cache; after that the tool works fully offline. There is no upload, no account, and no server step.
Press Remove Background once your image is loaded. The tool first downloads the segmentation model on first use, then runs inference and applies the predicted mask as the alpha channel of the output PNG.
The Download PNG button saves the masked image at the original resolution with full alpha. Soft hair and fur edges keep partial transparency from the predicted mask, so the cut-out drops cleanly onto a new background.
Entirely in your browser. The image is decoded into a canvas, the segmentation model runs on WebGPU when available or falls back to WASM, and the masked PNG is generated locally. Nothing is uploaded.
The segmentation model weights (around 180MB) are fetched from Hugging Face on first use and then cached by the browser. Subsequent runs reuse the cached weights and start much faster.
Photos with a clear subject and reasonable contrast against the background give the cleanest mask. Hair, fur, transparent objects, and very busy backgrounds may leave soft edges or small artefacts because this is a one-pass model.
If WebGPU is not available the tool falls back to WebAssembly, which is several times slower. Large images also take longer because the model processes the full resolution. Try resizing the image down before uploading if speed matters more than maximum detail.