Show HN: A tool to translate images across languages | Hacker News Reader

Show HN: A tool to translate images across languages

blog.imgtranslate.com

68 pointshbornfree20 comments

Do you happen to know about project naptha? It actually seems more advanced than what you have right now because it can detect text regions too.

For example in https://projectnaptha.com/, right click on the text on any of the demo images to erase, translate, or rewrite it.

Both your solution and their's produce blurry artifacts when it fails to inpaint/erase the text correctly. I wonder how you'd would sell this to marketing agencies etc if it can't produce pixel perfect results? Isn't it easier for them to just edit the text from the source photoshop file instead?

I wouldn't be at all surprised if this IS using project naphta behind the scene. Since it's pretty much the only open source solution that does this in js. Wrapping this in a pay for play service and not providing a github to the HN community, meh.

It isn't. ANd as mentioned in the blog post, we want to discuss how it works in the next post and open source as much as we can. We're built on open-source stuff after all. :)

looking forward to that post.

Yes we do know of Project Naptha. Infact I personally used it a couple of years ago. imgtranslate also has a text region identification module (through clustering mechanisms) but I have to admit it's not that great at the moment. So I've hidden it.

As far as inpainting goes, we should soon be switching to GANs which produce structurally more accurate and less blurry results.

The problem with photoshop is that - You need a lof of fonts installed. - It doesn't always render the non-latin fonts correctly - It's expensive

Hello everyone! The creator of imgtranslate here. Happy to discuss anything on tech and product. The current stack uses a good dose of traditional(OpenCV) and neural implementations. And we're actively pursuing LSPs(Language Service Providers) and marketing agencies to try it out.

Interesting, I can have an LSP try this. So, if it's not OCR, what is it? What is it doing?

It does OCR ofcourse. Pipeline is roughly:

OCR -> Text property identification -> Inpainting -> Region averaging -> Translation -> Fitting

Please have the LSP try it out and see if offers some value.

How are you going to compete with Google Translate?

By integrating the tool to professional CAT tools we're guessing. Businesses almost always never buy machine translated output

Great stuff, congrats!

Just a small (but important because is the hero image) correction from a spanish native speaker: the earth is female, so it should be "Hay que protegerla", ending in "a".

Thank you for the feedback! Will fix it right away

Nice idea! Why not just put pricing on the site instead of asking people to email for it?

I understand it's a deeply technical product and you probably just built an MVP (congrats!). But definitely think about distribution and pricing upfront. It will force you to decide who is your customer and help adjust the website copy accordingly.

E.g. You could be selling to developers (via API) or content creators directly but your current headline "Translate images across languages, easily" is not strong enough for either. Appeal to your customers pain point (saving time, effort etc.)

There's a small bullet that indicates it's a web app so I'm guessing it's aimed at end users directly. If that's the case, forget HN. Write more targeted copy, put some actual product images, add simple no-brainer pricing (it could even be free to begin with) and share it on some content creators / marketeers group on FB. Get the ball rolling, let the market decide. Good luck!

We haven't decided on the pricing nor implemented it in code. That's the next thing to do. Probably we'll roll out both bundles ($x for y images) or subscriptions.

As for the copy, you're right. We need a lot of improvement on targeting. We thought LSPs and marketers are going to be main customers. Have to refine there. Our first idea was to do an API but we deferred it as the output still needs a pair of human eyes to approve.

Thank you for the feedback and wishes. Really appreciate it!

I work with marketing agencies which has huge requirement for i18n & translations. I think there is definitely a market out there that would be interested in something like this. I have seen content creators creating different sets of images with translated text(takes many hours to manually do this right now). I am not sure how you are planning to offer this product/service, I would recommend offering it as an API which your clients can call from their applications.

> takes many hours to manually do this right now

This was also the original problem we set out to solve. Thank you for letting us know that more people share the same problem. Unfortunately we haven't yet gotten it into the hands of the marketing agencies.

If you are ok with it, can you send us a 'hello' mail at team{at}imgtranslate.com so that we can discuss a little more with you.

Cheers

Kind of a cool project and looking forward to seeing where it goes.

I was a little disappointed by what it actually was, because I was under the impression it would be somekind of OCR, that would translate text from an image ex. if you took a photo of a sign in one language you could translate it to another language by the photo you took.

What you're describing sounds like the Google Translate camera feature(erstwhile WordLens). If you haven't tried it out, you should. Works quite well on photos taken on phone for signboards etc.

We are little particular about what to translate and how to faithfully reproduce it. That's the main differentiation.

Hi,

Does this work with any Indian and how accurate is the translation with Indian languages.

Hi praveen, This works for these Indian languages as target language - Hindi, Punjabi, Bengali, Telugu, Tamil, Kannada and translation accuracy is as per Google Translate. If you are talking about Indian Languages as source image language, that is something we are working on to add.

Also, the main aim of imgtranslate is to ease the process of image editing, with a human intervention for translation verification.

PS: I'm also from imgtranslate team.

Reply on news.ycombinator.com