Vision \ image analysis

[razvanab]

Does this extension support vision \ image analysis?

[Shishir435]

Does this extension support vision \ image analysis?

Not at the moment. I explored adding vision support, but couldn’t find a reliable way to integrate OCR or image-processing libraries within Chrome extension constraints.

If you have suggestions or have implemented something similar, I’d love to hear them and will gladly experiment with possible approaches.

[razvanab]

I did something like that with vibe coding since I am not very good at this and maybe it will help.

Here it is:
background.js
content.js
manifest.json
popup.css
popup.html
popup.js
result.html
result.js
sidebar.css

[Shishir435]

I did something like that with vibe coding since I am not very good at this and maybe it will help.

Here it is: background.js content.js manifest.json popup.css popup.html popup.js result.html result.js sidebar.css

Thanks for sharing this, appreciate you taking the time to put together a working example and linking all the files.

Your approach works because it relies on OpenAI’s vision-capable API endpoint, where image understanding and OCR are handled entirely server-side. In that setup, the Chrome extension is mainly responsible for capturing the image and forwarding it to the API.

At the moment, Ollama Client is intentionally focused on local-only models running via Ollama. Most Ollama-supported models today are text-only, and Chrome extensions also have limitations around using native OCR or heavier image-processing libraries directly (especially in service workers / WASM contexts). Because of that, the same approach isn’t directly transferable right now.

I am actively working toward supporting external vendors like OpenAI, Anthropic, and Google, but that’s a larger effort and will take some time. Once vendor support lands, vision features like this become much more practical to integrate.

That said, if Ollama adds stronger vision-model support or if there’s a clean, local OCR setup that works reliably inside a Chrome extension, I’m happy to experiment and revisit this.

[razvanab]

The extension was made with Ollama in mind using REST API, so yes, it is compatible with OpenAI’s vision-capable API endpoint too.

[Shishir435]

The extension was made with Ollama in mind using REST API, so yes, it is compatible with OpenAI’s vision-capable API endpoint too.

While the extension uses a REST-style interface, it’s designed specifically for local, offline-first usage with Ollama. I’m intentionally avoiding cloud-based vision APIs (like OpenAI’s) to keep everything client-side and private.

Once reliable local vision/OCR support is feasible with Ollama or within Chrome extension constraints, I’m happy to revisit this.

[razvanab]

I understand.
Thanks.