I'm moving from Axxonnext, as they made it impossible for Joe Average to have their software.
I figured I'd try Agent DVR and I have it configured generally to my liking. Just one issue. It doesn't "Detect" as often as it should. If I walk into my garage, it sees me. If I take the same path, it consistently doesn't see me. There are many examples of it not Detecting. I've tried to make detection super sensitive and it just doesn't see me. All Detections are filter through Project AI AI object recognition (YOLOv5, full precision, large model) before recording, so I'd much prefer a false positive Detection.
Example configuration for a hikvision camera, where I use a 2688x1520 stream:
Detector: Simple
Sensitivity: 5-100
Gain: 40 (Gain confuses me a little. Defaults to 10, but AI says it's a multiplier of a default of 1).
Analyzer: CNT (only choice available)
Frame size: Medium (have tried large)
Max Object: 3
Detect Interval: 100
Track Interval: 100
Pixel Stability: 1-40
Tracking Timeout: 3
Movement Timeout: 3
It's triggering like crazy on bugs, but will miss my car leaving the garage. The camera in question is on my garage, pointed down the driveway. It'll consistently Detects the car entering, but not leaving.
I still have Axxonnext running on a less powerful machine (that I want to shut down forever), it doesn't have this issue. As a sanity check, I installed the trial version of Blue Iris in a less powerful VM and it doesn't have this problem. Except for this problem, I'm really liking Agent DVR. It gives a better end user experience than either Axxonnext or Blue Iris.
Thoughts?
I figured I'd try Agent DVR and I have it configured generally to my liking. Just one issue. It doesn't "Detect" as often as it should. If I walk into my garage, it sees me. If I take the same path, it consistently doesn't see me. There are many examples of it not Detecting. I've tried to make detection super sensitive and it just doesn't see me. All Detections are filter through Project AI AI object recognition (YOLOv5, full precision, large model) before recording, so I'd much prefer a false positive Detection.
Example configuration for a hikvision camera, where I use a 2688x1520 stream:
Detector: Simple
Sensitivity: 5-100
Gain: 40 (Gain confuses me a little. Defaults to 10, but AI says it's a multiplier of a default of 1).
Analyzer: CNT (only choice available)
Frame size: Medium (have tried large)
Max Object: 3
Detect Interval: 100
Track Interval: 100
Pixel Stability: 1-40
Tracking Timeout: 3
Movement Timeout: 3
It's triggering like crazy on bugs, but will miss my car leaving the garage. The camera in question is on my garage, pointed down the driveway. It'll consistently Detects the car entering, but not leaving.
I still have Axxonnext running on a less powerful machine (that I want to shut down forever), it doesn't have this issue. As a sanity check, I installed the trial version of Blue Iris in a less powerful VM and it doesn't have this problem. Except for this problem, I'm really liking Agent DVR. It gives a better end user experience than either Axxonnext or Blue Iris.
Thoughts?