Those really clever folks over at the Google Science labs, have created artificial intelligence software, which can describe the contents of photographs in a far more accurate manner than ever before. The Internet giant have said that as well as making it far simpler to search for specific images, the software may be used to assist people who are visually impaired understand pictures better.
The machine-learning software that has been developed by Google used two separate neural networks ; the first one deals with image recognition and the second one deals with natural language processing. If you are not aware, Neural networking is a computational model, which mimics some of the same architecture that is found in the human brain. Neural network systems have a series of interconnected neurones that are capable of taking information from a number of different sources and are also capable of learning.
Back in 2012 Google researchers had created some image-recognition software. They showed it 10 million images that were taken from YouTube videos. After only three days the programme had taught itself how to identify images of cats.
The neural network developed by Google was the work of four scientists; Samy Bengio Oriol Dumitru, Alexander Toshev and Erhan Vinyals. On the Google Research blog they wrote, “A picture may be worth a thousands words…But sometimes it’s the words that are the most useful – so it’s important we figure out ways to translate from images to words automatically and accurately.”
Who knows where this will take us? But one thing is for sure, when this system is fully integrated into the Google search engine, then finding things will become a whole lot easier.
[Image via tjournal.ru]