Orangeinks

Zerox’ Image Search Algorithm will Tell you if Photos are Good or Bad

 

Zerox’ Aesthetic Quality Image Search uses and algorithm that divides photos into two very straightforward categories, photos that suck and photos that doesn’t suck. If you think of it, when you are looking for a decent photo in the internet and don’t really have an eye for photos, what really will matter to you is that if is it any good or not. Zerox’ image search project will help you figure out if the photos that you are looking at are any good or not. This will be useful when there’s nobody else around to ask for an opinion.

The engine specializes in different subjects, like beaches, portraits and skies. The photos go through an image categorization engine that understands the content of the images (no texts and tags are required for the engine to process the images). The algorithm uses different parameters to evaluate the photos according to a particular subject matter. Then, the aesthetics rating engine separates the images with good and bad aesthetic elements.

Aside from telling bad and good photos apart, Aesthetic Quality also offers a few tips on how to capture images that are worthy of their “Good Images” section.

Here’s an example of the photography tip:

“A good beach image contains high cliffs, rugged coastlines, eroded landforms, pounding seas. Fast shutter speed (higher than 1/250) will stop the motion of the waves and freeze sea spray. Slow shutter speed will blur the waves and soften the seascape.

Add human elements and a sense of scale.

Head for piers and rocky promontories to include the sea in the foreground.

A polarizing filter will often improve the colour and contrast around the water.

Ref: Lonely Planet’s Guide To Travel Photography

 

Tags: Aesthetics, Image Search, Photos, search, Zerox

Google Image Swirl Finds Similar Images and Organizes them for You

Google Image Swirl is an experimental feature of Google for image search. This new feature combines technology and visualization to organize image search results based on their visual and semantic similarities. They are presented in an intuitive explanatory interface.

This tool can easily resolve an ambiguous query visually. For example, if you do a search and used the keyword “Jaguar,” the results page may return images of a cat combined with images of a popular car brand. This is also true for the beetle and apple keywords. You can also explore a concept from different visual perspectives like viewing certain structures on different point of views in an organized manner.

What Google Image Swirl does is combine a variety of image similarity features with additional metadata about the images to build a hierarchy of clusters of image search results. This is pretty much the explanation on how it works. But it really looks and feels better when you’re trying it out for yourself rather than reading about it. The interface is very self explanatory and user friendly.

This tool/feature has a true potential to make Google image search better and a lot easier. Instead of browsing pages after pages of image search results, you can easily find what you are looking for by just viewing the top image on a particular cluster or group of images.

Tags: Google, Google Labs, Image Search, Image Swirl

TinEye: The Reverse Image Search Engine

TinEye is a new search engine that exactly does the opposite of how you normally search for images. Instead of using keywords to search for images and hoping that the one who uploaded them wasn’t too lazy to tag them, you can do an image search using another image. Things just got a little easier, well, sort of.

TinEye may be used to find similar images to the ones you have. You can upload an image (or paste an image URL) and TinEye will do an image comparison using their image identification technology similar to the ones that you only get to see on CSI. Pretty cool huh! The search engine will display results similar to your images. As of now, their index is still small but it is continuing to grow and may become extremely useful in the future.

When you submit an image to be searched, TinEye creates a unique and compact digital signature or ‘fingerprint’ for it, then compares this fingerprint to every other image in our index to retrieve matches. TinEye can even find a partial fingerprint match.

TinEye does not typically find similar images (i.e. a different image with the same subject matter); it finds exact matches including those that have been cropped, edited or resized.

I believe that the TinEye search engine has a lot of potential and they are doing a great job with the image identification technology. I just wished that users are also allowed to do images searches the old fashion way. This will add more functionality to it by allowing the users to search for tagged images and use the image identification system to find similar images. I’m looking forward to the development of the software and perhaps develop it to search with keywords (like we normally do) and recognize the images through those keywords even if those images aren’t tagged.

iPhone users can also use TinEye to search for music through images. Just take a snapshot of the album art or grab them from somewhere on the web and let TinEye’s image identification system to do the work for you. It will give you links for that album on iTunes, allmusic.com, YouTube, and Wikipedia. No need to type the artist, album name or song title.

Whether you are doing research or having a little fun playing image detective, TinEye is worth checking out.

Tags: Image Search, iphone, Music Search, Reverse Image, search-engine, TinEye