For my BIT 330 class, I am doing my term project on Apple Inc. For this, I must analyze every aspect of the company and all of it's products. I like to include pictures of their products, and therefore today's class about image search was very helpful. Given that I must constantly update information for my project (which takes a ton of time), searching for good images is the least of my worries and is therefore something that I don't want to have to spend a lot of time on.
New Aluminum MacBook
One recent news article that I covered was Apple's media event October 14th, when Apple announced their new MacBook family. I wrote a big piece on this event, and I found myself looking for pictures of the new laptops. However, searching Google Images I couldn't find any. All of the images were of the old MacBooks or some other Apple product. I tried a couple different searches, such as "Macbook aluminum" and "new MacBook". I literally found 2 pictures (neither very good) after 15 minutes of searching. Even now (nearly 3 weeks after the event) there still aren't any new MacBook pictures from these searches.
Flickr is an excellent resources for high-quality user-generated pictures. Using Flickr, I did an image search for "new MacBook aluminum" and almost every result that came back was some variation of the new MacBook. Of the first 24 results, 18 were of the new MacBook aluminum and I found 8 that could be useful for my project. This is compared to Google finding 0 images of the new MacBook. Another thing to note is the quality of Flickr images is much, much better than those of Google.
Google News Search
Both Google News and Yahoo News provide pictures with their news updates. This can be beneficial if you are searching for images of something related to the news. Given that I wanted pictures from Apple's media event, this was the perfect tool for me. One thing to note is that while Flickr provides user-generated images, the images produced by Google News are much more formulaic. I did a Google News Search for "MacBook Aluminum" which brought back 250 results, many of which were awesome:
A Pixolu Search
Now that I had found a bunch of useful new MacBook pictures, I decided to search for some generally useful pictures of Apple Inc. I did this using Pixolu, which is the coolest image search feature I've ever seen. One quick note: When you do an image search in Pixolu, the web address does not change, so therefore I cannot provide you with links of the specific searches I did (you will have to do them yourself).
Start by going to Pixolu.de. On the home page, you can tell Pixolu how many images you want (ranging from 100 to 300) and where you want it to search (Google, Flickr, and Yahoo). After you enter your initial search into Pixolu (I searched "Apple computer"), it brings up all the images on one page. One nice thing that Pixolu does for you is it organizes your results by color. For my image search on Apple, it put all the white background images up top and grouped the other images by section (red background, dark background). From here, you can zoom in and scroll through the images. This is another neat feature because all of your images are on one page, and therefore instead of hitting next to see more images, you can easily move around the page zooming in and out.
What makes Pixolu stand out from the rest of image search engines is that Pixolu will specialize your search further for you. If you want to find additional images, you can select and drag the images on the screen which are most applicable for your search. Then hit the NEXT button and Pixolu will search for "semantically similar" images. This is an amazing feature, and offers a lot of potential! For my Apple search, I selected images of Apple's logo (the apple symbol) and hit NEXT. It worked flawlessly, bringing back all kinds of different Apple logos.
While my second search worked well, sometimes Pixolu has difficult refining the search for you. This is because Pixolu is still in the beta version and therefore does have technical difficulties. When we did this exercise in class, no one could get their semantic search to work because the site was having trouble loading it. Currently, the home page for Pixolu says:
Unfortunately, due to the high number of request we receive, Pixolu may be very slow and may not be able to return semantically similar images. Please come back in a few days.
Regardless, it seems that the inital search always works and therefore Pixolu is a very helpful resource.