Why Kerberos.Io over Zoneminder and Shinobi
I did a post not long ago about why I ditched Shinobi. But, truthfully, it mostly boiled down to the bug causing the massive files, in a feature I don't have in Kerberos and don't really need.
So, I felt it was best to take another stab at it and this time include the one piece of software most people looking at this sort of thing would be considering; Zoneminder.
As important as anything else, one needs to understand my particular requirements. If yours are different, it may make sense to draw a different conclusion on this topic:
It also allows me to monitor remotely via Portainer, and makes redeployment elsewhere a breeze.
All of the solutions had Docker images. But, ZoneMinder's was purely community driven and Shinobi advised against it (probably related to my exploding snapshot issue). So, I would say Kerberos won here.
I use TensorFlow within HA to process my camera feeds because it works, and generally works better than what ships with most smart cameras. I'm willing to cut it out if the software I'm using does better. Shinobi had great, highly configurable image detection. Kerberos is so-so, but it can push a message to MQTT which can trigger TensorFlow to process, and the timing is good enough. ZoneMinder I didn't get that far with. I will blame complexity here. Shinobi and Kerberos win here.
I'm recording locally, which means my disk space is being eaten up by whatever I decide on. The most important thing to me is getting videos of possible detections. Not a full on stream of every second of every day. Shinobi wins here. I can easily setup both. Kerberos doesn't seem to have a continuous recording mode at all though. So, that might kill it for others. Just, not for me.
Lightweight is where the competition falls apart. Both ZoneMinder and Shinobi wanted to foist full on R-DBMS solutions on me. I don't like that. Even with multiple cameras I would likely never need anything more than SQLite. And I suspect that the same is probably true of the vast majority of their user case. Kerberos uses almost no RAM or CPU unless something is actually happening. And even then it is lower than the others. My server is still one of my daily use PCs.
Reliable goes to Kerberos and ZoneMinder. I didn't use ZoneMinder as much as the others. But, it also didn't explode my server like Shinobi so I'll give it points.
User friendly goes to Kerberos and Shinobi. This is subjective. But ZoneMinder reminds me an early MySpace page with CCTV baked in. It seems more sterile and business grade. Which is fine by many. Not to my tastes. I like Shinobi's UI the best if I had to choose.
And so there you have it. As you can see, while all 3 solutions are decent. Given my goals, Kerberos comes out on top. An official Docker image that provides what I need with no extra requirements bloating out my system that has so far run without any hitches and integrated with HA sufficiently to get me accurate updates when there is something worth looking at going on.
So, I felt it was best to take another stab at it and this time include the one piece of software most people looking at this sort of thing would be considering; Zoneminder.
As important as anything else, one needs to understand my particular requirements. If yours are different, it may make sense to draw a different conclusion on this topic:
- Docker Support.
- A means of integrating with Home Assistant, and/or reliable motion and image detection.
- Recording events separately from camera feed (in fact I don't need a recording of the camera feed at all).
- Lightweight.
- Reliable.
- User Friendly.
It also allows me to monitor remotely via Portainer, and makes redeployment elsewhere a breeze.
All of the solutions had Docker images. But, ZoneMinder's was purely community driven and Shinobi advised against it (probably related to my exploding snapshot issue). So, I would say Kerberos won here.
I use TensorFlow within HA to process my camera feeds because it works, and generally works better than what ships with most smart cameras. I'm willing to cut it out if the software I'm using does better. Shinobi had great, highly configurable image detection. Kerberos is so-so, but it can push a message to MQTT which can trigger TensorFlow to process, and the timing is good enough. ZoneMinder I didn't get that far with. I will blame complexity here. Shinobi and Kerberos win here.
I'm recording locally, which means my disk space is being eaten up by whatever I decide on. The most important thing to me is getting videos of possible detections. Not a full on stream of every second of every day. Shinobi wins here. I can easily setup both. Kerberos doesn't seem to have a continuous recording mode at all though. So, that might kill it for others. Just, not for me.
Lightweight is where the competition falls apart. Both ZoneMinder and Shinobi wanted to foist full on R-DBMS solutions on me. I don't like that. Even with multiple cameras I would likely never need anything more than SQLite. And I suspect that the same is probably true of the vast majority of their user case. Kerberos uses almost no RAM or CPU unless something is actually happening. And even then it is lower than the others. My server is still one of my daily use PCs.
Reliable goes to Kerberos and ZoneMinder. I didn't use ZoneMinder as much as the others. But, it also didn't explode my server like Shinobi so I'll give it points.
User friendly goes to Kerberos and Shinobi. This is subjective. But ZoneMinder reminds me an early MySpace page with CCTV baked in. It seems more sterile and business grade. Which is fine by many. Not to my tastes. I like Shinobi's UI the best if I had to choose.
And so there you have it. As you can see, while all 3 solutions are decent. Given my goals, Kerberos comes out on top. An official Docker image that provides what I need with no extra requirements bloating out my system that has so far run without any hitches and integrated with HA sufficiently to get me accurate updates when there is something worth looking at going on.
Comments
Post a Comment