Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Computer vision holds the key to the smartest home, literally

Computer vision-based user data will improve our lives at home. There are various layers within a house that serve a specific purpose, and when they are combined, they create what we call a “home.” For example:

  • Some are meant to keep us secure: the door we lock, the alarm system and security cameras in our home
  • Some are meant to keep us safe and provide peace of mind: smoke alarms and baby monitors
  • Some are meant to keep us comfortable: climate control (thermometers), window blinds and lighting
  • Some are meant to entertain us: TVs, media devices and sound systems
  • Some are meant to maintain us: food and appliances in our kitchen, the covers on our bed and roof above our head

To reach new levels of security, safety, comfort and entertainment we need to embrace the idea of computer vision in the home, rather than shying away from the idea of sharing personal data. Computer vision allows computers/systems to understand what they “see” via digital images or video. When systems can detect and recognize objects, they can deliver smart actions according to what they were programed to do. One area that has effectively proven how computer vision can enhance our lives is the automotive space. Car systems that utilize computer vision to learn the driver behind the wheel and “see” the surrounding area can alert the driver when he starts swerving out of his lane.

Many consumers are already using computer vision on their smartphones and don’t even know it. Snapchat and Instagram are both using computer vision-tracking to detect facial features and place overlays (filters) in the right places.

Security, safety and comfort

Welcoming computer vision into your home and adding it to your connected devices will enable a new level of convenience to your everyday routine. Your front door will be able to “see” when you arrive and unlock the door for you, or remain locked when an unfamiliar person (face) approaches. The alarm systems will be smarter, able to detect specific members of the family (including age and gender) and who’s not. Taking it a step further, indoor security cameras will send an alert to your smartphone if an elderly family member or guest falls, or if a toddler is climbing up the stairs, on the countertop or anywhere that places the child in danger. Nest, Logitech and other smart home manufacturers have already started to offer these smart security features to consumers as a subscription-based premium service, or already embedded them in their newest products.

When your home can see and has the capability to learn, over time it will be able to adjust lighting and temperature to the number of people in any given room to ensure a comfortable environment and potentially save you money on your electric bill. Your fridge will be able to “see” when you are low on milk and eggs and send you an alert to make sure those items are added to your grocery list.


Imagine a cable box that greets you by name and turns on a tailored interface with the content you like. Imagine not having to watch ads that are not relevant to you in any way, and a TV that is smart enough to block inappropriate content/channels when it detects kids in the room. Imagine a system that notices your significant other is sitting next to you on the couch and suggests content that will be enjoyable to both of you. These are examples of how your home and media reacts and adjusts to you. You become a true master of your home and providing data gives you access to new capabilities and information previously unavailable to you.


Embedded computer vision is the answer to the ongoing data privacy debate, but consumers must first understand that there are two categories of computer vision systems in the home:

  1. Activities and services that aim to tailor the home to you (for example, adjusting the lights and temperature, suggesting content on TV to the people in the living room, etc.)
  2. Activities and services that you may want to monitor in addition to security system alerts (for example, sensing who is at the front door to unlock or sound an alarm, and monitoring your children or activities in home).

For the first category, there is no need for the computer vision systems to save, show or share any video and visual content. All that is needed for the systems to work is real-time data reporting to the relevant systems, and the output should be X people in room, age, gender. To customize media content, an output stating the relevant family members in front of media device is necessary, but no visuals are needed. This can and should be embedded, which means no videos or images are sent to the cloud, the camera shouldn’t save any imagery locally and the camera itself shouldn’t be connected to internet — all computer vision calculations should be done in real time on the device itself. If the above rules are met, the computer vision technology is embedded on the device and your concerns about privacy should be resolved because the hackability of the system is extremely low.

Regarding the second category of computer vision systems, you have a need and desire to monitor and see what is happening in your house, so the privacy protection measures with the home security cameras should be implemented here as well. Once again, the deeper understanding and seeing capabilities of these systems do not require processing in the cloud, internet connectivity of the sensor or saving of images. It is you that wants to view it, so the risk with these enhanced capabilities will be no different from the ones faced today in this sector.

All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.