SimCam AI Camera ReviewBy - 11/24/2018
Sorting through the troves of new gadgets at CES can be an overwhelming experience, so every year, CES bestows an Innovation Award upon a lucky few who stand out from the pack. One of the 2019 Innovation Award winners is a company called SimCam. The startup plans to launch an indoor/outdoor camera on Kickstarter that carries the company name. SimCam is an AI camera for home security and automation.
How is SimCam Innovative?
|SimCam||Lighthouse||Netatmo Welcome (Smart Indoor Camera)|
|Field of View||100° (360° Pan)||112°||130°|
|Built-in Siren||Sold Seperately|
|Person Recognition||Paid Plans Only|
|Recognition||Object Recognition||Gesture Recognition With Paid Plans||Listening Recognition (CO Alarm, Security Alarm, Siren, etc.)|
|Storage||Local||Cloud||Internal, Cloud, Dropbox, FTP|
|Integrations||Coming Soon: Alexa and Google Assistant||IFTTT, HomeKit, Google ASsistant|
In examining SimCam’s key innovations, we will also examine if the features are truly unique. Spoiler alert, they are not. What makes SimCam unique is that it performs tasks locally.
Instead of relying on the cloud, SimCam uses an Intel Movidius Myriad 2 AI processor. According to SimCam,
“Powered by the Intel Movidius Myriad 2 AI processor, SimCam can run 5 neural networks in parallel as model to provide far more intelligence than its default features.”
The first feature offered by SimCam is Activity Zones. Using the mobile app, you can create zones to tell your camera areas to monitor and ignore. For example, if you want to monitor your front yard, you might tell the camera to send you notifications if activity occurs on your porch while ignoring your street.
Another feature aimed at reducing false alarms is person detection. This helps SimCam distinguish people from bugs, trees, rain, and the dozens of other things that might trigger it.
Beyond detection, it also offers recognition. Again, this is nothing new. There are several cameras offering either person detection or recognition, including Nest Cam IQ, Netatmo cameras, Arlo cameras, and more. In this case, the fact that it’s processing locally isn’t even unique though it is rare. Lighthouse and Netatmo both offer person recognition, and they both perform at least some tasks locally. In fact, the three cameras have even more in common than just recognition. They also offer “unseen notifications”.
In order to use unseen notifications, you will have to teach SimCam recognizable faces. To do this, you will import photos of the people you want it to recognize and then tag them. The camera will then allow you to ignore “known people” or you can pick and choose who the camera should ignore and who it should alert for. You can also do the reverse and ask the camera to tell you if it
Person Classification and Action Recognition
SimCam pulls away from Netatmo Welcome and becomes more like Lighthouse as it uses Computer Vision to detect motion versus the use of a Passive Infrared sensor. Computer Vision will allow the camera to classify people and objects. For example, SimCam can tell you if the person it sees is an adult or a child. They also report that it can tell male from female, though I highly question this claim and it’s bound to be wrought with error.
Finally, they advertise action recognition as the ability to understand what a person is doing. For example, you might teach it to recognize a wave. This is, again, something that Lighthouse offers.
SimCam pulls away from the competition completely with its ability to rotate mechanically via a 360° pan. This ability is combined with a feature known as object tracking. When SimCam locks in on a person, for example, it can track that person’s movement. We’ve seen this feature in both high end and inexpensive cameras like Wyze Cam Pan. While motion tracking is a cool feature, we’ve always found it to be a little glitchy.
SimCam also plans to add another feature that we’ve never seen before: object tracking of actual objects. All of the tracking cameras we’ve tested thus far track movement. SimCam can also track objects. Using the app, you will train SimCam to track any object within the camera’s field of view. For example, you might train it to track your freshly baked cookies or your keys. If someone walks off with either, SimCam will follow the object’s movement.
As we’ve established, SimCam has more than enough features in place to reduce false alarms, but what happens when an event occurs?
When the camera detects activity, the first thing that happens is that it sends an alert with a snapshot via the mobile app. From the app, you can view a live feed, use two-way audio, or you can check the camera’s history. The camera history will show recorded clips which are proposed to be 15-seconds in length. Clips are recorded locally to an SD card. From the app, you can also set the card to keep continuous video recordings.
It’s hard to distinguish SimCam from others because it has so many features that overlap, so let’s break this down.
It’s like Wyze Cam Pan in that it can rotate and track objects, but that’s all the two have in common. It becomes more like Netatmo Welcome in that it can recognize people, offers local storage, and local processing, but it pulls away by offering advanced AI that is more akin to Lighthouse. Like Lighthouse, SimCam uses computer vision that allows it to analyze what it sees on a deeper level, but it pulls away from Lighthouse in that it processes all information, not just some, locally. Also, it’s slated to be considerably cheaper than Lighthouse. While SimCam hasn’t announced an official MSRP, their website indicates that the camera will be in the $100-$199 range. Also, there will be no monthly fee.
Bottom line, this is a camera for those who don’t want to depend on the cloud but want access to all of the new features usually reserved for cloud-based cameras. We will update this article once SimCam launches on Kickstarter with more information and an opinion. In the meantime, you can sign up for updates on their website.