AI Background Remover
Remove a portrait background in your browser and keep the original image dimensions. Download a transparent PNG or replace the background with a solid color.
Make your background perfect
Processing stays in your browser. AI libraries/models download from third-party hosts on first use.
Original pixel dimensions are retained. PNG file size may be larger than the uploaded image. No forced compression.
Original photo appears here
Processed photo appears here
Manual mask refinement
Keep brush restores parts of your photo. Erase brush removes unwanted areas. Brush changes are applied at original resolution.
For Person Only the largest foreground area is selected automatically. In Object/Product, click Pick Object then tap your subject in the image. Unsupported objects need manual editing.
The checkerboard represents transparency. Manual edits are saved into the downloadable image; click the image to paint. For best accuracy zoom the browser page before small strokes.
About this tool
Remove photo backgrounds online for free with on-device AI. Keep original pixel dimensions and download a transparent PNG or add a color background.
How to remove an image background online
- Upload a JPG, PNG or WebP image using the file picker or drag and drop.
- For a portrait choose Person / Portrait. For supported objects you can try the experimental General Object mode.
- Click Remove background. The browser downloads the model the first time.
- Preview the transparent result, or choose white, red, blue or a custom background.
- Download the full-dimension PNG. PNG preserves transparency; it does not guarantee the same file size or perfectly identical edge pixels.
Does it preserve original resolution?
Yes, the export canvas retains the original pixel width and height (within the 16MP browser safety cap). The segmentation mask is predicted at the AI model's working resolution and scaled to the original image, so fine edge detail may be imperfect.
Does it remove backgrounds from every object?
Portrait mode is intended for people. General Object mode is experimental and recognizes only categories supported by its underlying model. Complex products, intricate logos and objects outside its learned categories may not cut out accurately.
Free AI, no processing API
The AI processing runs on your device, so there is no per-photo paid processing API. You still need an internet connection to fetch browser AI code and the model. Speed depends on device memory, CPU and browser support; model inference and fine edges are not guaranteed.
Related image tools
Photo & Signature Resizer · Image Compressor · Image Resizer