Need an automatic image tagging API, any suggestions?
I think if images could be labeled automatically, Google would have abandoned the image labeler a long time ago. Unfortunately, computers have a lot of trouble understanding images.
Edit:
If you are interested in computer vision research have a look at CVPapers, especially Open Source Computer Vision Implementations. Automatic image labeling is far from being solved (unless you have a very specific/restricted set of topics).
Quote from The Google Guide from Tuesday March 13, 2007:
The words “Larry Page” and “Sergey Brin” appear near images of Eric Schmidt, or in image captions, or in links to those images. Google makes a guess that the words are related to the image. Google technology isn’t yet to the point where it can tell what’s in an image by looking at it directly.
HTH, don't get your hopes up too high.
P.S.: I hope you (or someone else) proves me wrong and shares it here with me ;-)
Edit2:
I just stumbled across the Voc 2010 Challenge, which, in my opinion, illustrates very well the current state of computer vision advances. In one of the challenges the contestants have to find an object (from a very limited set of objects) in the image and classify it. On the result page you can see, that one of the algorithms manages to classify air plane with a 93% accuracy, but "fails" at other categories.
That is just for the quest to find the "things", not even adjectives or emotions.
Check out https://imagga.com/ It has some impressive results. Also some wildly entertaining results... Thankfully all tags generated come with a confidence value, so you could always ignore anything less than a threshold (~15% for my use case). 12,000 images a month for free, not bad. If you have over 12,000 month images then just queue your requests.
It's not entirely clear whether you would like to define the tags to use yourself, or just let the software use a "common sense" universal set of tags about the objects shown etc.
Let's say you want to define your own set of tags — they can be about the year season a photo was taken in, a mood associated with the image (based on color scheme and depicted objects etc.), or something technical you need to distinguish (nudity, detail, background type etc.).
We can use machine learning for this! It's a branch of artificial intelligence that learns rules (like how to tag images — even very complicated rules) when we give it many examples of the images. So the main step for you is to gather a set of example images for each tag you want. Once you do this, for images you have two main options:
Use a deep learning framework which lets you apply neural networks on the problem. You will need to split your data to smaller parts, do quite a bit of coding and unless you have a lot of images, use a variety of tricks to get it learn your task well. Unless you are interested in research, caffe and TensorFlow are the ones to look at now (a year ago the recommendation was different, and a year from now it may be different again).
Use an online API, as you mention. But for the task where you want your own set of tasks, you don't have many options, as most services just do general classification - they sort your images based on what "daily life" objects they detect on the images (and sometimes special cases like NSFW, but often not on the sensitivity level you would like).
An option you have among web-based APIs is vize.it, which offers a web interface where you can upload and label your example images and it lets you train your own AI API which generates the tags you specified. So you are getting the best of both worlds. Unfortunately, it's not completely free, but the plan is fairly low-cost for small amount of images and you get a free sample at the begining (plus the training process is free too).
Disclaimer: I'm one of vize.it co-creators.