Nine years ago, I hired a security installer from Yelp who recommended 14 basic 4MP cameras from ENS Security (IP-5IR4002VFZ-W) and two PTZ cameras (IP-5PT96E2IR20X) for my home. The system was simple, lacking “smart” features, and its motion detection was unreliable, constantly triggered by shadows from my 50 palm trees swaying in the wind. However, it recorded decent-quality video with a two-week archive. The installer mounted a 32” CCTV monitor in my kitchen, making it easy to view the cameras.
Later, I added two Ring Floodlight Cams with solar panels, mounted on palm trees facing my house. These have performed exceptionally well out of the box, with highly accurate motion detection and virtually no false alerts. To be fair, their simpler scene—facing the house rather than the swaying trees—likely helps. They integrated seamlessly with my Amazon Alexa, announcing “Motion detected at the driveway/front door” when someone approached, prompting me to check my CCTV monitor displaying the 16 hardwired cameras. I still use these Ring cameras to this day, mostly for the Amazon Alexa alert but also as a backup in case my hardwired CCTV somehow fails.
Wanting more from my 16 cameras and not relying solely on Ring for motion alerts, I aimed for precise notifications when someone approached my front door (on the side of my house, covered by seven cameras) or entering my driveway or garage (covered by three to four cameras). I needed instant alerts without false triggers, tracking motion in near real-time as someone moved through areas like the driveway, corner, side, and front door. I joined the IPCamTalk forums and purchased Blue Iris, spending over 50 hours tweaking its motion settings to eliminate false alerts (this was before third-party AI). The challenge was balancing settings: too lax, and I got false alerts; too strict, and I missed real motion. Despite the effort, I enjoyed the process.
Learning from the forums that Andy from EmpireTech was a trusted vendor, I upgraded six key cameras to the 4MP Varifocal IPC-T54IR-ZE and one PTZ to the Dahua SD5A425XA-HNR with auto-tracking. I also replaced my outdated NVR with the Dahua NVR5432-16P-I. The new cameras’ quality was outstanding, and I appreciated not being limited to Internet Explorer for accessing their admin pages.
When CodeProject.ai emerged, I adopted it, achieving about 98% accurate motion alerts. However, the system occasionally crashed and required maintenance. Blue Iris alerts on my iPhone were unreliable, so I switched to PushOver, but the delay (5–10 seconds) due to AI confirmation and notification processing was too slow. I spent $800 on a MiniForum MS-01 PC with an i9-13900H as a dedicated Blue Iris machine, hoping it would resolve issues. While it consumed less power, it didn’t fully solve the delay problem.
I later switched to Deepstack, which added only 1–3 seconds to alert times and was relatively stable. However, delays sometimes reached 5–15 seconds, falling short of my goal for near-instant alerts. I accepted the system for several years but wasn’t fully satisfied.
A few weeks ago, I discovered a post about Unifi Protect and the AI Port device, which integrates existing cameras with Unifi’s advanced AI features. Since I already used a Unifi Dream Machine for my network, I added an 8TB drive and bought the AI Port for $200 to experiment. The results were astounding. Unifi Protect’s AI features surpassed Deepstack and Blue Iris, delivering instant iPhone alerts with zero false or missed positives. The user interface was vastly superior, requiring minimal setup. I added a Unifi G6 Instant 4K camera to test their “Advanced AI” features, like face recognition and license plate/vehicle searches across all cameras, which worked flawlessly out of the box. Unifi Protect’s modern interface, lack of need for port forwarding or VPNs, and plug-and-play setup won me over.
While exploring Unifi Protect, I found a forum thread about using the IVS (Intelligent Video Surveillance) feature on my IPC-T54IR-ZE cameras for accurate motion detection in Blue Iris without third-party AI like Codeproject. I had previously tried IVS and SMD but abandoned them due to false alerts. I tested IVS again on one camera, enabling ONVIF alerts and disabling Codeproject. Initially, false alerts persisted (e.g., car headlights triggering alerts without people or vehicles). A forum comment clarified that SMD and motion alerts must be disabled, leaving only IVS enabled with “person + vehicle” filters. After making these changes, I achieved a 100% success rate with no false alerts over the past 3 weeks of testing.
I also retested Blue Iris push alerts, which were faster than PushOver and linked directly to the alert video, resolving my earlier issues. Now, I have a fully functional Blue Iris setup with high-quality EmpireTech IPC-T54IR-ZE cameras, delivering instant, accurate alerts in complex, windy conditions. Additionally, I’m testing one camera with Unifi Protect via the AI Port and a Unifi G6 Instant 4K camera. Unifi’s platform offers advanced features like license plate recognition and face detection & following across multiple cams, which Blue Iris can’t easily match, with minimal configuration, and again, their UI is far superior to Blue Iris 5 in my personal opinion.
Summary: If you set up your Blue Iris system years ago, revisit its features or consider a fresh setup, as you may be missing improvements. For newer cameras with SMD and IVS, disable SMD and motion alerts, enable IVS with “person + vehicle” filters, and you might eliminate the need for third-party AI to reduce false alerts. Also, if you're looking for a brand new system that simply works out-of-the-box with a bunch of advanced AI features, consider the Unifi Protect system along with their G6 cameras as their UI is top-notch, and everything just works with no need to tinker with things. If I had to start all over from square one today, I would probably go with the Unifi G6 system, however I have zero regrets using my Blue Iris system which was working years before Unifi ever came out with a descent camera & NVR system. Hopefully BlueIris V6 comes out with a total revamp of their UI to keep up with modern CCTV systems as their old UI is functional but feels like it's from the 2010 era still.
Later, I added two Ring Floodlight Cams with solar panels, mounted on palm trees facing my house. These have performed exceptionally well out of the box, with highly accurate motion detection and virtually no false alerts. To be fair, their simpler scene—facing the house rather than the swaying trees—likely helps. They integrated seamlessly with my Amazon Alexa, announcing “Motion detected at the driveway/front door” when someone approached, prompting me to check my CCTV monitor displaying the 16 hardwired cameras. I still use these Ring cameras to this day, mostly for the Amazon Alexa alert but also as a backup in case my hardwired CCTV somehow fails.
Wanting more from my 16 cameras and not relying solely on Ring for motion alerts, I aimed for precise notifications when someone approached my front door (on the side of my house, covered by seven cameras) or entering my driveway or garage (covered by three to four cameras). I needed instant alerts without false triggers, tracking motion in near real-time as someone moved through areas like the driveway, corner, side, and front door. I joined the IPCamTalk forums and purchased Blue Iris, spending over 50 hours tweaking its motion settings to eliminate false alerts (this was before third-party AI). The challenge was balancing settings: too lax, and I got false alerts; too strict, and I missed real motion. Despite the effort, I enjoyed the process.
Learning from the forums that Andy from EmpireTech was a trusted vendor, I upgraded six key cameras to the 4MP Varifocal IPC-T54IR-ZE and one PTZ to the Dahua SD5A425XA-HNR with auto-tracking. I also replaced my outdated NVR with the Dahua NVR5432-16P-I. The new cameras’ quality was outstanding, and I appreciated not being limited to Internet Explorer for accessing their admin pages.
When CodeProject.ai emerged, I adopted it, achieving about 98% accurate motion alerts. However, the system occasionally crashed and required maintenance. Blue Iris alerts on my iPhone were unreliable, so I switched to PushOver, but the delay (5–10 seconds) due to AI confirmation and notification processing was too slow. I spent $800 on a MiniForum MS-01 PC with an i9-13900H as a dedicated Blue Iris machine, hoping it would resolve issues. While it consumed less power, it didn’t fully solve the delay problem.
I later switched to Deepstack, which added only 1–3 seconds to alert times and was relatively stable. However, delays sometimes reached 5–15 seconds, falling short of my goal for near-instant alerts. I accepted the system for several years but wasn’t fully satisfied.
A few weeks ago, I discovered a post about Unifi Protect and the AI Port device, which integrates existing cameras with Unifi’s advanced AI features. Since I already used a Unifi Dream Machine for my network, I added an 8TB drive and bought the AI Port for $200 to experiment. The results were astounding. Unifi Protect’s AI features surpassed Deepstack and Blue Iris, delivering instant iPhone alerts with zero false or missed positives. The user interface was vastly superior, requiring minimal setup. I added a Unifi G6 Instant 4K camera to test their “Advanced AI” features, like face recognition and license plate/vehicle searches across all cameras, which worked flawlessly out of the box. Unifi Protect’s modern interface, lack of need for port forwarding or VPNs, and plug-and-play setup won me over.
While exploring Unifi Protect, I found a forum thread about using the IVS (Intelligent Video Surveillance) feature on my IPC-T54IR-ZE cameras for accurate motion detection in Blue Iris without third-party AI like Codeproject. I had previously tried IVS and SMD but abandoned them due to false alerts. I tested IVS again on one camera, enabling ONVIF alerts and disabling Codeproject. Initially, false alerts persisted (e.g., car headlights triggering alerts without people or vehicles). A forum comment clarified that SMD and motion alerts must be disabled, leaving only IVS enabled with “person + vehicle” filters. After making these changes, I achieved a 100% success rate with no false alerts over the past 3 weeks of testing.
I also retested Blue Iris push alerts, which were faster than PushOver and linked directly to the alert video, resolving my earlier issues. Now, I have a fully functional Blue Iris setup with high-quality EmpireTech IPC-T54IR-ZE cameras, delivering instant, accurate alerts in complex, windy conditions. Additionally, I’m testing one camera with Unifi Protect via the AI Port and a Unifi G6 Instant 4K camera. Unifi’s platform offers advanced features like license plate recognition and face detection & following across multiple cams, which Blue Iris can’t easily match, with minimal configuration, and again, their UI is far superior to Blue Iris 5 in my personal opinion.
Summary: If you set up your Blue Iris system years ago, revisit its features or consider a fresh setup, as you may be missing improvements. For newer cameras with SMD and IVS, disable SMD and motion alerts, enable IVS with “person + vehicle” filters, and you might eliminate the need for third-party AI to reduce false alerts. Also, if you're looking for a brand new system that simply works out-of-the-box with a bunch of advanced AI features, consider the Unifi Protect system along with their G6 cameras as their UI is top-notch, and everything just works with no need to tinker with things. If I had to start all over from square one today, I would probably go with the Unifi G6 system, however I have zero regrets using my Blue Iris system which was working years before Unifi ever came out with a descent camera & NVR system. Hopefully BlueIris V6 comes out with a total revamp of their UI to keep up with modern CCTV systems as their old UI is functional but feels like it's from the 2010 era still.
