CanDoYa
All tools
Saved toolsYour favorites

Add to favorites

Saved in this browser only.
Private browser utility / Social & Creator

Free Image Caption Generator

English only

Runs entirely in your browser - no upload, no sign-up.

Live workspaceLocal processing

Choose a photo to generate three private AI descriptions.

Share this tool
image captioner / browser utility
01 / Overview

What is an image caption generator?

An image caption generator uses computer vision and language generation to turn a picture into a short written description. This tool creates three editable English drafts inside your browser. Your photo stays on your device while a public AI model supplies the visual and language knowledge.

02

How to use

  1. 01
    Choose an image

    Drop in one JPEG, PNG, or WebP up to 10 MB and 16 megapixels, or load the sample scene to explore the workflow.

  2. 02
    Generate captions

    Start the local model. The first run downloads public model files, then processes the image locally on your CPU.

  3. 03
    Review and copy

    Edit the three suggestions to correct context or details, then copy one caption or the complete numbered set.

03

Who it's for

  • Editors and bloggers drafting neutral captions before adding names, places, dates, or story context.
  • Accessibility teams creating a starting point for alt text that will receive a human accuracy and purpose check.
  • Photographers and archivists producing searchable description ideas for images that should not be uploaded to a captioning API.
  • Students and developers exploring browser-based image-to-text inference without installing Python or sending a photo to a server.

This captioner runs the ViT-GPT2 image-captioning model through Transformers.js 3.8.1 in a dedicated browser worker. It downloads roughly 250 MB of 8-bit model weights and processes the image through WebAssembly on your CPU. The model and runtime download separately from public hosts, but the selected image is processed locally.

ViT-GPT2 joins a Vision Transformer image encoder with a GPT-2 text decoder. It was trained for concise English descriptions rather than promotional social copy or complete accessibility decisions. Use its suggestions as a first draft, then correct missing context, visible text, names, identity, and details that only a person can verify.

FAQ

Is my image uploaded to a server?

No. The selected image is decoded and analyzed inside a browser worker on your device. The worker downloads the public Transformers.js runtime and ViT-GPT2 model from external hosts, but those requests do not contain your image. CanDoYa does not receive or store the source photo or generated captions.

Is this image caption generator free?

Yes. There is no account, payment, credit limit, or watermark. Your device supplies the computing power. The first run downloads large public runtime and model files, so it still uses your internet connection and may matter on a slow or metered plan.

What image limits apply?

The tool accepts one JPEG, PNG, or WebP image up to 10 MB and 16 megapixels, with each side at least 32 pixels. These limits reduce browser memory spikes. Animated images, GIF, HEIC, SVG, camera RAW files, and multi-image batches are not supported.

How large is the image captioning model?

The tool downloads roughly 250 MB of 8-bit model weights. Runtime and configuration files download separately. Browsers may cache these files for later runs, but storage cleanup or private browsing can remove them.

Can I use an AI caption as alt text?

Use it only as a draft. Good alt text depends on why the image appears on a page, not just what a model sees. Check the caption against the image, remove irrelevant detail, add important context, and leave the alt attribute empty when an image is purely decorative.

How accurate are the generated captions?

Accuracy varies with the scene. Clear photos of common objects and activities tend to work better than small details, diagrams, unusual viewpoints, specialist subjects, or images dominated by text. The model can omit objects or invent relationships, so compare every suggestion with the original before publishing it.

Does image captioning work without WebGPU?

Yes. This tool uses WebAssembly on the CPU for consistent browser compatibility, so WebGPU is not required. Processing stays on-device and can take several seconds depending on your hardware.

Can I run the image captioner offline?

A later run may work without another download after the browser caches the runtime and model, but offline availability is not guaranteed. Browsers can evict cached files, storage cleanup can remove them, and private browsing often discards downloaded model data when the session ends.