An international team of computer scientists has developed a new image-recognition software. They found that 256 to 1,024 bits of data were enough to identify the subject of an image. The researchers said this 'could lead to great advances in the automated identification of online images and, ultimately, provide a basis for computers to see like humans do.' As an example, they've stored about 13 million images picked on the Web and stored them in a searchable database of just 600 megabytes. The researchers added that using such small amounts of data per image makes it possible to search for similar pictures through millions of images on your PC in less than a second. But read more...
Let's see how the method works. As shown on the left, it is possible to represent images with a very small number of bits and still maintain the information needed for recognition. "Short binary codes might be enough for recognition. This figure shows images reconstructed using an increasing number of bits and a compression algorithm similar to JPEG. The number on the left represents the number of bits used to compress each image. Reconstruction is done by adding a sparsity prior on image derivatives, which reduces typical JPEG artifacts. Many images are recognizable when compressed to have around 256-1024 bits. (Credit: Torralba et al.)
This research work has been led by Antonio Torralba, an assistant professor at MIT 's Computer Science and Artificial Intelligence Laboratory. Torralba collaborated with two other assistant professors in computer science, Rob Fergus of the Courant Institute of Mathematical Sciences at New York University, and Yair Weiss of the Hebrew University in Jerusalem.
Here is a quote from Torralba about this project. "We're trying to find very short codes for images, so that if two images have a similar sequence [of numbers], they are probably similar--composed of roughly the same object, in roughly the same configuration. If one image has been identified with a caption or title, then other images that match its numerical code would likely show the same object (such as a car, tree, or person) and so the name associated with one picture can be transferred to the others. With very large amounts of images, even relatively simple algorithms are able to perform fairly well in identifying images this way."
But how can you retrieve images using such a small amount of data? The illustration on the left shows representative retrieval results. "Each row shows the input image and the 12 nearest neighbors using ground truth distance using the histograms of objects present on each image (Credit: Torralba et al.) Please note that other methods are employed.
This research work will be presented in June 2008 at the IEEE Computer Vision and Pattern Recognition conference (CVPR 2008) in Anchorage, Alaska. The researchers will present a technical paper named "Small Codes and Large Image Databases for Recognition." Here is the link to this paper (PDF format, 8 pages, 20.18 MB), from which the images seen here have been extracted.
And here is the beginning of the abstract. "The Internet contains billions of images, freely available online. Methods for efficiently searching this incredibly rich resource are vital for a large number of applications. These include object recognition, computer graphics, personal photo collections, online image search tools. In this paper, our goal is to develop efficient image search and scene matching techniques that are not only fast, but also require very little memory, enabling their use on standard hardware or even on handheld devices."
And if you want to learn more -- and have fun -- please visit the 80 Million Tiny Images project page set by Torralba.
Finally, here is the conclusion of the MIT news release. "Torralba stresses that the research is still preliminary and that there will always be problems with identifying the more-unusual subjects. It's similar to the way we recognize language, Torralba says. 'There are many words you hear very often, but no matter how long you have been living, there will always be one that you haven't heard before. You always need to be able to understand [something new] from one example.'"
Sources: David Chandler, MIT News Office, May 21, 2008; and various websites
You'll find related stories by following the links below.