Remove the background from an image
Planned as a local AI tool: segmentation running on your own device, not on a server.
Not available yet
This tool is on the roadmap. The page is here so the workflow, formats and related tools are documented — but nothing is processed and no result is produced.
Background removal is the single most requested image operation on the web, and it is also the one most commonly implemented by uploading your photo to somebody else’s GPU. KaiConvert’s plan is different: run the segmentation model in your browser through ONNX Runtime Web, using WebGPU where the device supports it and WebAssembly where it does not.
That design is why the tool is not shipping yet. A usable segmentation model is tens of megabytes, needs to be lazy-loaded, warmed up and run inside a Web Worker so the page stays responsive, and needs a graceful path on devices that can do none of that. Shipping a fake version in the meantime would be worse than shipping nothing.
What you get
On-device inference is the goal
The model runs in your browser, so the photo never needs to be uploaded to be segmented.
WebGPU with a WASM fallback
Accelerated where the hardware allows, still functional where it does not, disabled with a clear message where neither works.
Feeds straight into the editor
A cut-out is rarely the end of the job. The result will land in the same canvas as crop, resize and export.
How it works
- 1
Capability detection
KaiConvert checks for WebGPU, WebAssembly, OffscreenCanvas and available memory before offering the feature.
- 2
Lazy model load
The segmentation model is fetched only when you actually use the tool, then cached by the browser.
- 3
Worker inference
Inference runs in a Web Worker so the interface stays responsive, and the mask is composited onto the canvas.
Specifications
- This tool is not implemented yet. Nothing is processed, no image is uploaded, and no result is produced. The page documents the intended workflow and is excluded from search indexing until the feature actually works.
- Planned as Tier B (local AI): ONNX Runtime Web, WebGPU with a WebAssembly fallback, inference in a Web Worker.
Why use KaiConvert
- No upload for local tools
- Works offline once the page is loaded
- Cloud AI is always clearly labelled before you start