Everything You Need to Know about AI Security Cameras
Today we are going to talk about AI (artificial intelligence)-powered security cameras and their place in the home security ecosystem. Together we will discuss Artificial Intelligence and Computer Vision to see how these technologies are being implemented by security camera manufacturers and how they can improve the security of our home video surveillance systems.
What is an AI Security Camera?
An AI-powered security camera is a camera that utilizes intelligent video analytic software to interpret specific objects and behaviors in the captured video. This level of analysis happens automatically and can make video captured by an AI camera significantly more valuable than video captured by a regular camera. By comparison, an AI camera gives the user real-time security monitoring while your standard camera gives the user the ability to live stream events.
What can an AI Security Camera do?
An AI-powered security camera can detect multiple objects in the live footage, then classify and localize the objects with a bounding box. The camera can give you real-time actionable data such as alerts and video clips when a predetermined criterion is met.
The following are some of the most common uses of AI in terms of video surveillance.
Person detection is one of the main uses for AI and Computer Vision in a security camera. With Person Detection, you can tell if a moving object is a person based off of models that were taught to the AI. Person detection can allow a user to be alerted to the presence or absence of a person in each area and can be applied to both sides when someone enters or leaves the area.
Facial recognition is the next most heavily used feature of the AI. Facial Recognition takes person detection one step further and can allow the end-user to identify specific people by facial characteristics. For facial recognition to work, faces must be loaded into a database. Cameras like SimCam make this happen by simply uploading a picture of the face to the device. Other Cameras like the Nest Cam IQ can learn faces on the fly, but for this to work, these cameras must be trained on an ongoing basis. It should be noted that the technology is not perfect. Factors such as skin tone and lighting can impact the reliability of facial recognition.
The uses of facial recognition technology can range from alerting a user to the arrival (or departure) of a family member, alerting users to the presence of an unfamiliar face to access control using “Face ID” type technology.
Vehicle detection, as the name implies, allows users to be alerted to vehicles entering or exiting an area. You may not think this feature is valuable compared to facial recognition or person detection, but that is most likely because you have not used it. Vehicle detection has many uses both in the residential environment and in the commercial and industrial space as well.
Security cameras with animal detection capability can help people identify when a pet enters or leaves an area and what type of animal it is.
Animal detection does have its limitations currently. You will notice the name of this feature is animal detection and not pet detection. Currently, the technology is not there to identify specific pets or if that pet belongs to you. Perhaps in the next few years, we could expect a facial recognition technology exclusively made for pets.
Traditionally, motion can be tracked by the changes in pixel state around something. With A.I.-enabled motion tracking, a security camera can lock onto an object and track its motion in a series of frames. Object tracking combined with PTZ (pan, tilt, and zoom) functionality of some advanced security cameras allows for better tracking of objects outside the camera’s field of view and can help in identifying people, places and things.
A.I. cannot only be trained to recognize certain objects but also can read almost all kinds of behaviors such as sleeping, drinking, laughing, cry and so on. Law enforcement officers used A.I. surveillance cameras to predict crimes before they happen. Business owners adopted A.I. in detecting shoplifting. Google Clips, a smart life-logging camera by Google, can automatically capture a photo or a video clip whenever you smile. The limits of behaviors that A.I. can recognize are only in the minds of engineers.
Smart Home Control
A.I. security camera can trigger your smart home devices or services based on the recognition of human activities. Typical uses could be turning on the lights when the camera detects a person and off the lights when there is no person around. Smart Home Control through the use of an AI Camera can be very powerful and can utilize almost anything that the camera can offer. Like turning on the appropriate lights based off the face it recognizes or allowing access to that same person while denying access to a face it does not recognize.
How can AI Vision Understand So Many Things?
With everything discussed, you are probably wondering how so much can be understood by these cameras? The answer to that is twofold. First, there is the sheer volume of highly skilled engineers that work on artificial intelligence pushing to advance the technology. Second, there are massive amounts of data that is being fed into these devices. With every bit of data gathered and interpreted, the AI can become smarter assuming engineers can adjust and retrain algorithms along the way. We are far from self-learning “sentient” artificial intelligence as right now human involvement is required to teach the AI what it knows but it is because of this commitment to AI that end users can enjoy the many benefits.
Why do You Need an AI-powered Security Camera?
Maybe you have now realized that you ‘want’ an AI Camera, but are still on the fence. Here are some reasons why you ‘need’ AI in your cameras.
- Increased motion detection because AI is more responsive than looking at changes in pixels
- Decreased false alarms because AI is smarter than PIR sensors
- Precise monitoring because AI can work faster saving you time from watching endless video
These are all the core benefits of AI-Powered Security Cameras and they should appeal to anyone.
Cloud AI Versus Local AI?
Cloud versus Local is a hot button debate right now in just about everything technology. People argue on both sides of the fence and claim that one is better than the other. Let’s start with a definition of both then I will give you the points that should allow you to make your own decision.
- Cloud AI happens on a remote server somewhere. Data is sent from the device to this server to be analyzed and is then sent back down to the device for execution.
- Local AI happens right on the device itself taking out any transmission time to and from the cloud. Local AI does, however, require more computing power and thus better hardware to execute.
The Downside to Cloud-based AI
Cloud-based AI can be very attractive as you can use hardware that does not have the same specs as local AI, but this comes with some tradeoffs:
- Higher data usage as large amounts of data need to transfer to and from the cloud
- Slower response time as you need to wait for this data transmission which can be affected by current internet speeds
- Privacy as once the data is sent to the cloud it could be used for other things
- Cost of data storage as most manufacturers are not going to cover this cost unless of course, they are selling your data to a third-party to cover it
As you can see Cloud-Based AI has its drawbacks and these all need to be considered. However, some companies are doing an excellent job executing cloud-based security systems.
Best Cloud-Based AI Security Cameras
Nest offers AI in the Nest Hello Video Doorbell and the Nest Cam IQ line of cameras, which offers person detection, facial recognition, and motion tracking. These features come at $10 per month or $100 per year for the first camera and $5 per month or $50 per year for each additional camera.
Arlo was originally released as a battery-powered camera and was very popular because of its range and battery life. In 2018, Arlo introduced Arlo Smart which added AI functionality to its cameras via a firmware upgrade. This comes with costs ranging from $2.99 per month per camera up to $14.99 per month per camera.
First released in 2017, Lighthouse is an indoor security camera with sophisticated AI capabilities. It can identify persons and pets, recognize faces, and even tell if it’s an adult or a kid. The camera comes with a 3D sensor so it cannot be fooled by a 2D image. Unfortunately, Lighthouse has shut down its business in 2018.
Why is Local AI better for security applications?
Local AI, also called on-device AI, in most cases is hands-down better than cloud-based AI. For most it will come down to privacy as local AI happens locally your data stays in your control but there are many more advantages like:
- Faster response
- Enhanced reliability
- No long-term contracts
- Keeps your data in your control
One thing to keep in mind however when it comes to local execution of AI is that hardware is much more important. You need hardware and components that can handle the massive amounts of data required for the AI to function properly and timely.
Best Local AI Security Cameras
When it comes to Local AI, we have two players in the space.
SimCam is a very versatile security camera with a set of unique embedded AI features. In addition to person and facial recognition, SimCam can alert you if an object is missing, and follow a moving person in 360 degrees. Because SimCam processes its AI locally, there is no subscription charge.
Netatmo is a French company specializing in smart home devices. In 2016, Netatmo introduced A.I. features in its Netatmo Presence and Netatmo Welcome security cameras. Like SimCam, Netatmo processes its AI locally so there are no service costs for the “Smart” features.
Having used many of these products, both local and cloud-based for me the choice was simple, local was the way to go.
The Future of the AI-Powered Smart Home
With current technology, everything that is called “AI” or “machine learning” still requires the interaction of humans to train the systems. As systems continue to be trained and technology continues to advance, we will soon move to a place where the “AI” can train itself. It will be able to make decisions on its own with a high level of accuracy.
AI security cameras are just the beginning when you consider the potential of an AI-Powered Smart Home. Consider for a minute the possibility of your home tailoring its behavior to the person (or people) who are in the home. You could set lighting, water temperature, and ambient entertainment and climate settings to the person. This is a future that is rapidly approaching and one that I am excited about. AI is the gateway into this world and will help to usher in the technology and user comfort level required to our homes and lives to transform from connected to truly smart. Are you ready for truly smart home and the world?