Mexican engineer made a "blind camera" that "listens" to objects (3 photos + 1 video)

The camera is guided by the sounds around and using artificial intelligence creates pictures.





Mexican engineer Diego Trujillo Pisanti developed original digital camera, devoid of the usual lens. Instead of using a sensor to capture light, the device is guided by sound picture of the environment, creating an image on it basis.



In general terms, an unusual device called Blind Camera used as a normal camera. It is enough to point it at the object, and press the shutter button. Installed on the front of the device a horn that is used to amplify the sounds being picked up.

After processing the data, a special AI algorithm generates a finished image. The module is used as a hardware platform Raspberry Pi 3B, equipped with a small screen that acts as viewfinder. The latter also informs the user about the process processing and displays the generated image.



According to Pisanti, he created a custom artificial neural network (ANN) using Python 3 and trained it with videos recorded around Mexico City. The model has been created by analyzing each frame and accompanying it with the last second audio sequence. This helped create sound and video associations that the system can use to create images. Now all created apparatus, “photographs” are largely based on images cities.