Performance:

This camera provides nice crisp and clear images during the day as you can see below, but despite having a passive infrared sensor for motion detection it does not have an infrared light source for night vision. This makes the images at night very noisy in low light to all out black if there is no ambient light. In the night image below the porch light provide enough light to have an idea of what is going on but certainly any advanced features like facial recognition will not work.



Initially I was really excited to try out this facial recognition and set alerts to only be for unknown faces. But the implementation of this just misses the mark for me. What I have found is that almost all events are labeled as motion. I think this is due to the cameras location; it simply is too far away from where people come into the frame so motion is detected first and rarely are faces detected even if they walk right up to the camera in the clip. Obviously part of this is due to my particular situation but it is further compounded by the design of the camera. Unlike traditional cameras that have a base that allows you to fine tune your elevation, this camera only sits flat. So ultimately you are very limited in how you can mount it. They also didn’t include any wall mounting holes or brackets and with the large battery this is simply too heavy to use double stick tape, in my opinion. They also missed the opportunity to have a threaded camera mount on the bottom, at least with that you would have options readily available for just about any mounting location you could come up with.

So mounting issues aside, I also found that even after repeated training of people faces from different angles it rarely recognized them, I still just get a notification for face detected. I also found some peculiar things while trying to train the faces. Sometimes while you are combining faces to a person the app will add another “person” without a title. But backing out of the app and returning shows that this phantom entry is not actually there. Also on at least one occurrence the camera found a “face”, which happened to be a wreath my wife was hanging. As you can see in the image below it selected part of the wreath as a face but not my wife’s which was also in the frame.



In the end I could never really figure out what made it detect a face. I did try moving this camera around to see if that helped, but it was still a mixed bag. In the event log shown below you can see my face clearly and the side of my son’s face as we sat on the floor within 10 feet of the camera. That event was labeled as motion but it did detect my son’s face just prior to that.



I also tried bringing this camera to my office to test its battery back-up and offline capabilities. Again this fell a bit short for me. Obviously without an internet connection you won’t get any notifications but it would have been nice if there was alert when the device lost its connection to the cloud and when it re-established the connection. This way you have a time frame of events to review. I was also disappointed that while there were dozens of people passing by this camera it rarely recognized a face. And when it did detect a face, these faces never appeared in the screen to train them when the camera returned back to the home network.



Sound: The clarity of the sound in the video clips is on par with all other IP cams we have tested as well as the clarity of the two way audio. Also, while I did get sound notifications, I would have liked to see a threshold setting for this. I found it was a bit too sensitive at times and often would alert when the door was closed.

Event recordings: The Ooma Butterfleye does buffer some video and they claim it has the ability to capture 5 seconds before the event occurred so you don’t miss anything. While I am not sure if I always got 5 seconds before the event, I would confirm that I had a few seconds of recording before say the motion was seen. But what I don’t like about this camera as well as many other security cams, is they only record clips. So if someone kicks in my door, I will have a clip that is say 5 seconds before the event and maybe about 15 seconds after. So even if the intruder is in front of the camera for 10 minutes I may only have a few clips of them. Unlike some other cameras that will continue to record until the motion stops which would give you the best chance of catching the persons face. I could understand if this was done in battery mode to conserve power, but this is just how it operates regardless of power source.