MIT's MemNet Algorithm Can Predict How Memorable An Image Is
MemNet, an algorithm developed by researchers from MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL). The algorithm uses Artificial Intelligence to predict how memorable or forgettable an image is nearly as accurately as humans do. For each image, MemNet produces a heat map of images showing which parts of the image are most memorable.
The technology is now available through an online demo that anyone can try. Their technology is able to predict whether people are likely to remember or forget an image, and that it can do it to near-human levels. Researchers are planning to launch a mobile application for the same.
Researchers have tested the algorithm on tens of thousands of images from several different data sets and are very satisfied with its result. The algorithm went head-to-head with humans by predicting how memorable a new, never-before-seen image was.
As it turned out, the MemNet algorithm performed 30 percent better than existing algorithms and was within a few percentage points of the performance of the average human.
This algorithm is expected to assist in online image posting and digital image processing for social media sites. Aditya Khosla, who was lead author on a related paper is scheduled to present the paper in Chile this week at the International Conference on Computer Vision.
Google, Facebook, Xerox and and Office of Naval Research are among the organizations that are supporting the research along with the National Science Foundation, the McGovern Institute Neurotechnology Program and the MIT Big Data Initiative at CSAIL.
Though MemNet is not better than humans at the task of measuring memorability of images, researchers expect that it will get better over time.
Source: <a href="https://news.mit.edu/2015/csail-deep-learning-algorithm-predicts-photo-memorability-near-human-levels-1215" target="_blank" rel="nofollow noopener noreferrer">Deep-learning algorithm predicts photos’ memorability at “near-human” levels | MIT News | Massachusetts Institute of Technology</a> | Technical Analysis:#-Link-Snipped-#
The technology is now available through an online demo that anyone can try. Their technology is able to predict whether people are likely to remember or forget an image, and that it can do it to near-human levels. Researchers are planning to launch a mobile application for the same.
As it turned out, the MemNet algorithm performed 30 percent better than existing algorithms and was within a few percentage points of the performance of the average human.
This algorithm is expected to assist in online image posting and digital image processing for social media sites. Aditya Khosla, who was lead author on a related paper is scheduled to present the paper in Chile this week at the International Conference on Computer Vision.
Google, Facebook, Xerox and and Office of Naval Research are among the organizations that are supporting the research along with the National Science Foundation, the McGovern Institute Neurotechnology Program and the MIT Big Data Initiative at CSAIL.
Though MemNet is not better than humans at the task of measuring memorability of images, researchers expect that it will get better over time.
Source: <a href="https://news.mit.edu/2015/csail-deep-learning-algorithm-predicts-photo-memorability-near-human-levels-1215" target="_blank" rel="nofollow noopener noreferrer">Deep-learning algorithm predicts photos’ memorability at “near-human” levels | MIT News | Massachusetts Institute of Technology</a> | Technical Analysis:#-Link-Snipped-#
0