A new study by the Stanford University’s Computational Imaging Lab suggests that using a technology that passes a laser beam through a small hole in the wall knows what’s in the room, but a month later, This time, the MIT team announced that the technology could be used to find out what’s going on without seeing the solution.
When light enters a dark room, the naked eye can tell whether something is moving and stationary in the dark, or whether something is round and angled. However, in some areas of the light, it will be almost dark and you will not see anything. However, a new study using neural networks is that it is possible to know how many people are working on what kind of work only from an image of a wall.
In the MIT study, first, a white wall is photographed using a high-resolution camera with excellent performance in low light, and the shadow of the movement of people is erased from the video, and furniture, etc. are stationary, and a room condition is assumed.
Do the same thing on another wall in another room, but this time set up a situation where a specific person performs a specific action. By learning this information to the neural network, they were able to reveal 94.4% of the people in the room and 97.3% of specific movements.
If you can get a lot of information from the shadows, you can think of closing the curtains and shutters right away. Of course, it is almost impossible for all approaches using neural networks to learn all human movements in the system. However, it is possible to teach them to be specific about certain illegal activities. Meanwhile, being able to identify whether, for example, is a bundle of bribes in a briefcase handed over to you by someone is another matter.
In addition, it is said that it will not work completely because it becomes difficult to distinguish between the shadows caused by human movement and the moving shadows due to the blinking of light in a state where lights such as candles or TV are blinking. In addition, even in a dark room, accuracy is reduced due to camera performance limitations. Related information can be found here.