Internet of Things

Hitachi releases new video analytics software for better facial and object recognition while protecting privacy

Hitachi Video Analytics analyzes video data without compromising individual privacy, yet offers better options for public safety.

istock-482112104.jpg
Image: iStock/alice-photo

Hitachi is applying advanced analytics to video data to improve how it's used for public safety, traffic analysis, people counting, and queue detection, as well as object, license plate, and facial recognition.

The advancement is part of Hitachi's smart city and public safety portfolio that uses Hitachi Video Analytics (HVA) to access the untapped potential of the video footage that is collected in many cities.

HVA can improve public safety without compromising individual privacy, which has been an ongoing concern as cities add more video cameras to the streets.

"One of the main restraints is privacy concerns. When people see video cameras going up, they immediately think about privacy," said Justin Bean, director of smart city solutions marketing for Hitachi.

SEE: Smart cities: 6 essential technologies (TechRepublic)

Privacy has been a market weakness for video analysis. It's a major concern for organizations deploying video solutions in public areas because it leaves companies at risk of lawsuits, as people want to protect their individual privacy. Some vendors have pixelated the faces of people in frames to hide their identity, but this can still leave a significant portion of their body exposed, and they might still be identified by clothing, tattoos, or birthmarks.

To avoid this, HVA will be automatically redacting a person's entire body, so there is no personally identifying information revealed. The video footage can still be used for insights and alerts without compromising privacy. If raw footage does need to be reviewed, only authorized users with a unique chip card and PIN to authenticate a person's identity can access it. This reduces risk and liability for companies using video security and monitoring systems on premise.

HVA uses intelligent object detection to overcome pixel limitation, so there are fewer false positives in any video being used to detect intrusions and objects, or count people and vehicles.

"The method that we've used for analyzing the video is beyond just the pixel color and pixel movements. What it does is it actually reconstructs a 3D environment from a 2D image and it adds the fourth dimension, time, to analyze particles instead of pixels. It recognizes an object based on variables in three dimensional space and provides you with better accuracy because it knows there is a person who is 6 feet tall 100 feet away, and there is a person five feet away who is also 6 feet tall," said Anthony Caputo, project manager for Hitachi.

There's reduced "noise" such as rain or snow in the final image with HVA, because it can detect this, and mask it from the final view. This also reduces false positives. In addition, shaky images can also be stabilized and analyzed in 4D.

There's a valuable use for HVA in airports and train stations, Bean said.

"With HVA, [police] can detect a suspicious object that's been left somewhere. With that you can look back into the camera feeds and see the suspect that dropped that object off, and you have an image of that person. Next you can do live face matching from the video feed with [the] suspect identified. You're able to find where this person is much more rapidly than before and apprehend them and keep them from doing more potential harm," Bean said. "And you can find out where they came from. With this we can recognize that there was a car that dropped them off, and see which vehicle it was and then search the city traffic cameras to be able to apprehend the second suspect. It's a proactive approach to public safety and security."

It's also useful in retail stores, because retailers can determine where to place ads to get the most impressions by watching how customers shop in a store, and what causes them to stop and walk into a store instead of passing it by. "We have new metrics that can help these retailers fine-tune their business models," Bean said. "You can see what is working and what isn't instead of going on gut reaction."

The top 3 takeaways for TechRepublic readers

  1. Hitachi Video Analytics (HVA) improves video analysis for smart cities.
  2. HVA can improve public safety without compromising individual privacy, which has been an ongoing concern as cities add more video cameras.
  3. HVA reduces noise from snow or rain in the final video image to help eliminate false positives.

Also see

About Teena Maddox

Teena Maddox is a Senior Writer at TechRepublic, covering hardware devices, IoT, smart cities and wearables. She ties together the style and substance of tech. Teena has spent 20-plus years writing business and features for publications including Peo...

Editor's Picks

Free Newsletters, In your Inbox