I am using streams from my NVR and not from the cameras themselves so I think the fps and quality is lower then it would be. The only thing I can think that would help is better camera angle that are not so high up, and maybe just higher quality stream. However excluding the false 'people' positives is not going to catch the people it thinks were objects. It only allows you to create one include or one exclude box so excluding two trees might call for a complicated include box - but it does let you create a polygon and add corners/vertex not sure what the limit on the corners is. Either to exclude things like a tree or the neighbors driveway. I haven't tried the mobile app.Īllot of my false alarms could be fixed with boundary boxes.
Sighthound camera feed skyipcam 1620 software#
The remote web view while nice doesn't show the boundary boxes on the objects / people like the software does which would have been nice otherwise its fine.
![sighthound camera feed skyipcam 1620 sighthound camera feed skyipcam 1620](https://ssl.cdn-redfin.com/photo/rent/79f1bbcc-20f7-49b0-8f16-64e65239af9f/mbpaddedwide/genMid.12_1.jpg)
The entire reason to use it over other motion detecting is to find the people and not sift though hundreds of random movement, but if you have to look though all the object detection to find the rest of the people it defeats the purpose. Now this might have something to do with my high cameras and the angle I dont know.
![sighthound camera feed skyipcam 1620 sighthound camera feed skyipcam 1620](https://i.ytimg.com/vi/18M8vN1h8qQ/maxresdefault.jpg)
So if your just monitoring the people detection it seems that its not actually going to catch all or even most of the people. The other two were detected as just 'objects' - the same objects that have hundreds of false positives from stuff moving in the wind. And of these 3 tests it only detected one of the three has a 'person'. The worst thing is not the false 'people' positives but that I tested all 3 cameras by walking in their path. 3 being tree movement and 3 were the neighbors car. Out of the 7 "people" detected, 6 were false alarms. Most of the object detections were just tree/bush movement from the wind or at night light reflected from someone driving past (street nowhere near camera). These video doorbells will ring the touchpads and IP phones inside the house.
![sighthound camera feed skyipcam 1620 sighthound camera feed skyipcam 1620](https://ssl.cdn-redfin.com/photo/rent/70406427-0653-4dfe-b9eb-81ed7e53c55c/bigphoto/0_1.jpg)
My plan is to mount video doorbells at both entrypoints. From one day it made 230 clips from object and person detection. We do have cameras and home alarm system in place, but the goal of this system is to be able to clearly deny or allow access to people on our property. The 'person' (not object) detection has had a few false alarms. The detection isn't working nearly as well as they make it seem.įirst is the false alarms. The remote web access is really clean and simple as well. Probably the most simple one I've ever used - a little too simple for my taste but really good for non computer tech to use. To enable this platform in your installation, add the following to your configuration.I've been testing it with my cameras, the app is really simple to use. Note that by default the component will not automatically scan images, but requires you to call the image_processing.scan service, e.g., using an automation triggered by motion. If save_timestamped_file is configured as true, then the annotated image is saved with a file name that includes the time of detection. The saved image shows the bounding box around detected people and can be displayed on the Home Assistant front end using a Local File camera, and used in notifications.
![sighthound camera feed skyipcam 1620 sighthound camera feed skyipcam 1620](https://i0.wp.com/twistedsoulmusic.org/wp-content/uploads/2022/08/Album-Front-Panel-e1660826310244.jpg)
If save_file_folder is configured, on each new detection of a person, an annotated image with the name sighthound_latest.jpg is saved in the configured folder if it doesn’t already exist, and overwritten if it does exist. The event data includes the entity_id of the image processing entity firing the event, and the bounding box around the detected person. For each person detected, an sighthound.person_detected event is fired. This integration adds an image processing entity where the state of the entity is the number of people detected in an image.
Sighthound camera feed skyipcam 1620 pro#
If you need more processing per month you will need to sign up for a production account (i.e., a Basic or Pro account). The Sighthound Developer tier (free for non-commercial use) allows 5000 images to be processed per month. Detect people in camera images using Sighthound Cloud. Detect people in camera images using Sighthound Cloud.