rodrhern

OMG secure networks argument is so outdated. In that case internet is useless, online banks, payments, are damned, please someone warn them. Internet penetration in USA is about 88.5%. New trends of services trends to user internet, nest is streaming, other cameras has also introduced P2P streaming, dahua and hikvision DVR/NVR and cameras allows remote control and remote monitoring, tomcctv please tell them that they are just wasting their money on those useless features. If I follow your advice, I have to stay at home watching over my cameras, because exposing them to internet is so insecure. tomcctv "beach watch", really???, do you have something smart to say???, you are making yourself look like a real fool.

When I bought my first camera with motion detection, at the end of the first day of usage, my inbox was full of spam: every bug, bird, tree, or anything moving in the scene was a new picture in my inbox. Useless feature for outdoor monitoring. When is used for indoor, false alarms are related with shadows, or pets. NEST deal with this problem sending only 1 motion detection message every 30 minutes, they only reduced the spam but did not solve the problem. If you are installing a camera to get information to prevent or react a against a crime, an outdoor camera is preventive, while an indoor camera is reactive (burglars are inside already). A non monitored camera is not preventive and is not reactive, its only function is to be able to record in HD how burglars get into my home and robbed my stuffs. There are a lot of footage videos on youtube. Anyone with a little knowledge of Computer Vision would know that person detection in uncontrolled environments is a very very challenging task, and it's impossible to think in 0.0 error, just FYI an average person would make a 5.1 error rate in classification tasks, while Deep Learning has already outperformed that score. Is very clear that you are monitoring cameras, and not just drawing boxes into the recorded video and you proved that not in the near future but now, Computer Vision is doing a very decent surveillance.

You are really going to have troubles with the motion detection and outdoor cameras. Any shadow, tree, cloud, will be detected as motion and very quickly you are going to be full of spam, checkout http://get.sentinel.watch for an smart monitoring, not sure if exists other alternatives. About the QNAP NAS, have made a quick google search, seems that the software supports cameras, but I can't tell. https://www.qnap.com/en/tutorial/con_show.php?op=showone&cid=122

Why not buying a already designed NVR, Dahua NVR are not expensive and already designed for that task. There are also custom made software to monitor your NVR channels and display the videos. Anyway, if you are not rendering 3d images, or transcoding video, I think you don't really need a GPU.

Agree!!! You need a wireless setup, more than a standalone wireless camera. Basically an IP camera and a wireless Point-to-Point wireless link between the camera to the base unit. Since there is a long distance you should try with a directional antenna instead of an omnidirectinoal antenna, in order to gain some dBi. maybe this can help https://airlink.ubnt.com/#/ also be aware that rain, snow, can rapidly degrade you wireless link budget.

Hello Guys, new member here!!! I Have been working on Computer Vision based video monitoring and surveillance. Basically avoid humans to monitor our cameras and let the task to deep learning algorithms, so you can filter motion, pets, cars, lights, shadows, etc. and alert only when there is a real threat. This forum has been very helpful for finishing the develop of our system, now I hope I can help back!