BrainChip launches new AI software that can recognize faces at ultra-low resolution

According to foreign media reports, on Wednesday, BrainChip became a new player in the security field, launching a new artificial intelligence software called "BrainChip Studio", which they believe will change the way law enforcement and intelligence agencies use video surveillance to detect and distinguish faces. Let's follow the mobile phone portable editor to learn about the relevant content.

BrainChip launches new AI software that can recognize faces at ultra-low resolution

Not only does the software enable law enforcement to search vast amounts of video recordings faster and more efficiently than most traditional facial recognition solutions, it also does so without some of the typical limitations.

For example, BrainChip Studio can detect and distinguish faces from low-resolution video (just 24x24 pixel images) instead of high-definition video. The software can also use video recorded in poor lighting conditions.

In a recent trial, BrainChip’s software was able to accurately detect, extract and classify more than 500,000 videos recorded by eight different cameras. In another trial, it processed 36 hours of footage in two hours and extracted 150,000 facial images.

"We take a live or recorded video, detect faces, and then for each face we detect it to find we create a spiking neural network model," explains Robert Beachler, BrainChip's senior vice president of marketing and business development, explaining how the software works. "Then we take the next image, detect the face and compare it to the previous model. If it's a match, we know it's the same person in the field of view and we can track someone through the entire camera's field of view and then we basically group those images into tracks. In any given camera, there's going to be a person crossing it and then they become a track and there might be 20 images of their face that we've captured and we do that for all the faces in the camera's field of view."

He continued: "We keep a cache of these faces so we can track them as they cross our field of view. Once we gather this information, we use these spiking models to catalog the images into a database. We then take the person we're trying to find a suspect image of and run it through our spiking neural network to create a model and compare it to other images we've acquired previously. Once we've picked up on a face that we think is a high probability suspect, we look at all the images that have been tracked by that person. All of this information is put into a ranking, and the person you're looking for, we classify it as the most likely match."

BrainChip Studio’s other AI solutions utilize “Spike Neural Network” technology, a type of computing that more closely resembles how the human brain functions by mimicking the functions of neurons and synapses.

Neurons are made up of synapses that connect to other neurons, and the neuron itself is an integrated function of all the different inputs, which then decides whether the information should be passed on to the downstream neuron. This transmission of information is called spiking, and it is really a bioelectronic process, which is a signal of a certain strength passing through the synapse. "The way the network is trained is to strengthen or inhibit the connection of the synapse, changing the threshold function at which the neuron fires. So when you apply the same features of the human brain to software, the computer becomes better at recognizing patterns in video images."

The above is an introduction to the new artificial intelligence software launched by BrainChip in portable mobile phones, which can recognize faces at ultra-low resolution. If you want to know more related information, please pay more attention to eeworld. eeworld Electronic Engineering will provide you with more complete, detailed and updated information.