Back

Anki Image AutoGrabber Add-on: Gauging interest w/ potential release next week

#1
I have a working proof of concept of an add-on that would really benefit this community, and any Anki based community for that matter. But I need everyone's help to make it happen. 

tldr; Take text, grab image, add to card.

Anyone who donates $5 gets it forever (Will send it out within 24hrs). Releasing publicly in one week if donations reach $100.
Sample deck and proof screenshots are below.

[Donate via Paypal](https://www.paypal.com/cgi-bin/webscr?cm...5N5FJN8JJN)


**What is Anki-Image?**

Anki-Image is an add-on that takes any chunk of text (sentence, single vocab word, kanji fields, etc.) and downloads an image for the card and puts it on the flashcard.


**Why?**

 Context

* By getting the most relevant image associated with whatever you are studying (vocab, sentence, etc.), you can get more insight into what people think of when they say that word. This will help you visualize and remember even abstract concepts. After using this for some time now I've already noticed that words I thought I understood, I have a different insight to them now, and words that I had trouble picking up before are now easier to remember.

Automation

* Manually doing this for your Core 2k or Genki Annihilation deck would take you a decade. Each card fetch takes around 2s to complete currently with the add-on -- not bad, for the amount of work its doing.

**How**

Magic. Python. Google. More Python. Very similar to other add-ons I've released like Japanese, Heisig, or Phonetics add-on, the plugin will give you a new menu option in the Anki CardBrowser's Edit menu. This will allow you to select cards, press a button, and voila - images are now on your card. Behind the scenes it's connecting to a couple servers that take the keyword you are looking for and returns an image that is then downloaded and put in your media storage via Anki.

**Technical Challenges**

Anki makes life really difficult to do something like this. Besides being limited to a *subset* of platform inconsistent python2 methods, besides terrible unicode handling - essential to foreign characters, and besides by default not having a library that can handle https, the real killer for Anki add-on development is that you can't easily include dependencies to fix all the broken fore-mentioned things. Believe me I tried. But after hours of work - finally reaching a spot where everything *should have* been working, I was downloading corrupted images. So I had to find a new solution. Also one that would easily port to Anki 2.1 when it came out. So I introduced a server to offload all the difficult things to. That server now handles the hard parts so the Anki add-on can stay simplistic and be easily portable.

**Known Issues (donations will go to help with this too)**

* 95% success rate, I have tested this on my windows machine (and OSX) and for my 200,000 card collection of all types, I get around 95% success. I'm still working on making that 100%.



**Why ask for donations**

For the above listed reasons and depending how this goes, server costs. Besides the time already invested in this, I'm the one who is going to have to trouble shoot and work with you individually when things go wrong. I would be really grateful to the community if I could get just a small piece of compensation for the effort coming up to support this.


**But wait...**

What happens if you shut your server down or they go down temporarily?

* I have a super-duper, easy-peazy solution for this. Seriously, if you have the skills to sign-in to here, you won't have a problem with my solution. I'm not going to talk about it though unless it gets released publicly.

How can we trust you?

* I've been a part of the Japanese learning community for 3 years (as of next month ?). I've released a handful of other plugins, so I've got some experience with what I'm rambling on about. In the end, I'm just another passionate J-learner working towards a goal and trying to help other get there as easily as possible. 

* My other plugins: 

[Bulk-Generate Japanese Phonetics](https://ankiweb.net/shared/info/1459981298)

[RescheduleMe](https://ankiweb.net/shared/info/748123673)

[Mult-Field Japanese Support - JtoJ essential](https://ankiweb.net/shared/info/2062803440)

[Bulk-add Heisig's Keywords](https://ankiweb.net/shared/info/1855482626)

[Anki-Image](https://www.paypal.com/cgi-bin/webscr?cm...5N5FJN8JJN)


POC vocab deck: https://ankiweb.net/shared/info/199079496

POC images: http://imgur.com/a/lfqYf

**Donate**

[Donate via Paypal](https://www.paypal.com/cgi-bin/webscr?cm...5N5FJN8JJN)
Reply
#2
Sounds interesting. Is it just returning the first google images result? Sometimes the first result isn't helpful, but for the majority of times when the first result is good enough this would help.
Reply
#3
(2017-08-07, 6:12 pm)yogert909 Wrote: Sounds interesting.  Is it just returning the first google images result?  Sometimes the first result isn't helpful, but for the majority of times when the first result is good enough this would help.

Typically yes. It can handle other search engines - bing, tumblr, etc. but I've found google to be relevant (dare I say even useful), for even sentences around 80% of the time for my cases. Eventually it would be ideal to make some sort of retry logic if the image isn't relevant for you, but I'd hold off on that until the project takes off (or doesn't).
Reply
Thanksgiving Sale: 30% OFF Basic, Premium & Premium PLUS Subscriptions! (Nov 13 - 22)
JapanesePod101