Search Here

MIT Researchers have come up with an algorithm that can predict a photo's memorability as good as humans. This means that it will be able to identify faces, names, objects and scenes in photographs that are remembered by humans. The algorithm, called the MemNet, is based on deep-learning methodology. This technology can also be used by the politicians to create posters that will become unforgettable. This algorithm is expected to be used by the social media sites for online image posting and digital processing of an image. The paper about this study was revealed at the international conference on Computer Vision in Chile. LaMem is is the largest image-memorability dataset in the world. It has images with detailed metadata and annotations. There will also be a heat map that will reveals the parts of an image that are most memorable. "Understanding memorability can help us make systems to capture the most important information, or, conversely, to store information that humans will most likely forget," says CSAIL graduate student Aditya Khosla, who was lead author on a related paper. "It's like having an instant focus group that tells you how likely it is that someone will remember a visual message," as reported by MIT News.

According to the research team members, there are many ways to use this technology such as improving the content of the social media posts and ads, developing more effective teaching methods etc. The MIT research and its significance lies in the brain's ability to remember. The network is fed data that it rearranges to give more accurate predictions. The team then used its algorithms against humans by having the model foretell the parts of a never-seen-before image and how memorable it will be, reports Waltonian.