I received my IP8M-T2669EW-AI camera on 11/3/2021 and unboxed it. I checked what else was in the box, and it was the usual screws, cover for the ethernet cable connection, and documentation. I removed the camera and looked it over. I liked the metal turret that surrounds the round camera module, and it felt very solid. Since there were no surprises, I wasted no time mounting it on the front of my house. This was simple because I already had a camera mounted at that location and the wall mount was compatible with this camera. Just a few screws and the camera was working.
I logged into the camera and configured it to my preferences. I saw that the camera showed it was only capable of 20 frames per second (FPS) when the resolution was set to the maximum (4K). As with many others, I then read this was because of the AI features. I decided I wanted to evaluate the AI features later and turned them off. I then saw that the camera indeed supported 4K at 30 FPS. I was pleased with the clear picture and wide field of view, which I have come to expect from Amcrest cameras.
I checked how the camera looked once it was dark outside. I saw that anything outside the infrared LEDs was not visible. I logged into the camera, selected Camera > Configuration > BLC Mode > and selected BLC. I also selected General under Profile. I was then able to see much more around my house that was outside of the IR. I turned off the IR LEDs and it looked better than when they were on. This is in the settings under Camera > Configuration > IR Light. I also selected General for the Profile. Since IR was off, I decided to go to Camera > Configuration > Day & Night and I selected General under Profile and Color under Mode. Everything was now in color, though a bit muted with some graininess. I still found this better than black and white or when using IR. I do live in a tract house with streetlights and a lot of exterior lighting, but nothing more than a porch light from my house. Things may vary depending on your area.
I decided to try human detection, which I had only ever used on one other camera. I first tried to configure my email (Gmail) exactly per the instructions on the Amcrest website, but email tests failed. I finally got Gmail working and this was how:
- Configure your Gmail account to allow “less secure access” (https://support.amcrest.com/hc/en-us/articles/360001079832-How-To-Setup-Email-Alerts)
- Select Other as the email provider – do not select Gmail
- Enter smtp.gmail.com
- Select TLS, but change the port to 587
- Enter your Gmail address as the username
- Enter your Gmail password
- Enter your Gmail address as the sender
- Enter the title for the emails
- Enter the email address of the recipient and select the + sign so the email address is moved down to the box below
- Test it!
I did not have to do this with another Amcrest camera I had used, so Amcrest may need to check the firmware on this camera. I know it is a constant battle with Google to use Gmail with third party access, so this is not a criticism.
I turned on Human Detection in the camera settings (Event > Smart Motion Detection > check Enable). Soon after I was receiving email with photos of people walking, riding bikes, or even when they were barely visible in low light conditions. The AI certainly works well, but the only drawback was that a good number of photos did not have people visible in the still images (I was not using video clips). I did not do an exact count of photos with and without people, but it could be as high as 50% in my use. However, I am not saying someone had not been there a mere second before the photo was taken. If the camera sent three photos, one or two may have captured a photo of a person and the remaining photos would not have a person visible. I wondered if this may be related to the overhead of using AI and the drop in frames per second from 30 to 20 when at the maximum resolution. I decided to test this by dropping the camera resolution to 1080p and increasing the FPS to 30 while AI was still in use. I found that I still received a lot of still images without people in them, just as when the camera was at 4K resolution and 20 FPS. I randomly selected photos where no people were present at a particular time and then I checked the video footage for all cameras at that time. I did find that there were some false alerts with no people in the area around the time the still images were taken and emailed by the camera. Even with some false positives, I found the AI reliable enough that if I received an email alert of human detection, though nobody could be seen in the photo, I would still check video at the time of the photo if the camera was in a secure area. I also cannot claim my settings were perfect, so your results may be better or worse than mine.
I had never tried to save photos or video to my NAS, so gave it a try. I created an account for this camera on my NAS, but I could not get files to save to the NAS. I’m sure it was possible, but I instead chose to use FTP. This was simple and it worked right away. The root directory of for FTP was the “surveillance” directory I created for the camera. When I configured the camera FTP settings, I left the Remote Directory blank, and this created a directory within the FTP root directory on the NAS in the name of the camera. This was perfect since I could have multiple cameras saving to the same FTP root directory. I chose to have the photos deleted after three days, which I later confirmed did work.
My final feature to test was the IVS Tripwire (Event > IVS). I had never used this feature before and I really like it! You can draw a line and specify if you want to be notified if anything crossed the line. You can even select the direction travel. I drew a line close to my house and selected either direction of travel. I then received notifications with photos when something crossed the tripwire (mostly me). As with human detection, I did receive photos with nothing obvious present that could have crossed the tripwire. I selected one of the alerts and then looked at video for the same time period. I saw that it was me bringing in the trash. I was only briefly visible on video on 11/9/2021 between 14:17:21 (H:M:S) and 14:20:31 because I was passing through the very corner of the camera field of view. I received a tripwire alert for that date with the timestamp 14:19:49. I am please the tripwire managed to detect such a minimal target.
As I briefly mentioned before, the picture looks quite good. This is my highest resolution camera yet and I did not know I needed this resolution. I have a 1080p camera with a 25X optical zoom and it looks great when I zoom in objects. The problem is that I cannot always watch that camera and actively zoom in on objects. Recording in 4K cannot come close the capability of a 25X optical zoom, but the resolution does allow me to enlarge recorded footage just a bit to get a closer look at something. The higher the resolution and video quality, the better you can zoom in on recorded footage. That is why I would recommend setting the recording quality to the maximum. I would not worry too much about how much storage space this will consume because this camera has H.265 compression and video files are quite small. Only recording when there is motion also saves a lot of space compared to constant recording.
This was the most I had experimented with Amcrest AI features, and I think it is worth the extra cost if you will use the features. This could be especially useful for a business where personnel would want to be notified of a tripwire activation, person, or vehicle being detected in a particular area.