search

sr99622 / libonvif

Onvif library with GUI implementation and built in YOLOX
GNU Lesser General Public License v2.1
173 stars 44 forks source link

Added alert Panel in Gui #89

Open mriamnobody opened 7 months ago

mriamnobody commented 7 months ago

Added alert panel in gui where user can enter BOT_ID and Chat_ID to get alerts of image of detected objected in opted in. The logic for getting images after prediction is work in progress. This is the UI part. This also contains the ability to save the images of detected objects locally.

mriamnobody commented 7 months ago

Also the ability to remember user selection/choice made in Alert panel is work in progress.

mriamnobody commented 7 months ago

There is a small issue related to placement of buttons:

onvif gui version 2 0 9 18-04-2024 08_05_39

sr99622 commented 7 months ago

Thank you so much for the pull request. I am not familiar with telegram, it looks like an interesting option that might be viable. At some point I would like add this type of feature, but it should be accompanied with the ability to access the system remotely. I haven't had time to look, but there should be some existing solutions that solve most of the problem, i.e. live streaming the camera output to a mobile device through a web interface.

I'm currently working on more extensive testing for the basic functionality of the program, and adding support for open vino. The intel graphics cards have been producing some pretty impressive results on ubuntu, but not so much on windows. I'm finding that the nvidia cards are not producing satisfactory results on windows either. There may be some tweaking that could bring the windows performance to an acceptable level, but it may end up that windows is just not the right platform for the system.

Testing gets rather involved, as the application is designed to run unattended for indefinite lengths of time without user intervention which presents some challenges. The open vino on ubuntu has had the best performance so far, a very stable solution. The nvidia solution on ubuntu is also quite stable, but does tend to drift a little over time with the cache building up slightly. Windows performance out of the box is less stable and not what I would consider to be acceptable for a wider audience. There may be some tweaks that could bring the performance up, but I haven't nailed it down just yet.

mriamnobody commented 7 months ago

Thank you so much for the pull request. I am not familiar with telegram, it looks like an interesting option that might be viable. At some point I would like add this type of feature, but it should be accompanied with the ability to access the system remotely. I haven't had time to look, but there should be some existing solutions that solve most of the problem, i.e. live streaming the camera output to a mobile device through a web interface.

I'm currently working on more extensive testing for the basic functionality of the program, and adding support for open vino. The intel graphics cards have been producing some pretty impressive results on ubuntu, but not so much on windows. I'm finding that the nvidia cards are not producing satisfactory results on windows either. There may be some tweaking that could bring the windows performance to an acceptable level, but it may end up that windows is just not the right platform for the system.

Testing gets rather involved, as the application is designed to run unattended for indefinite lengths of time without user intervention which presents some challenges. The open vino on ubuntu has had the best performance so far, a very stable solution. The nvidia solution on ubuntu is also quite stable, but does tend to drift a little over time with the cache building up slightly. Windows performance out of the box is less stable and not what I would consider to be acceptable for a wider audience. There may be some tweaks that could bring the performance up, but I haven't nailed it down just yet.

I'm glad I could of some help—it was really enjoyable for me as well! The ability to control the system remotely, live stream, and the feature I have PR'ed are three distinct functionalities. The ability to receive Telegram alerts when an object is detected is truly a lifesaver, as we can't sit all day monitoring our cameras.

The features of live streaming and remote access are indeed useful, and I would love to see them integrated into this great tool, but they differ from the feature I have proposed in this pull request.

I was also not very familiar with many of Telegram's features, including those used in the 'Alert' feature. However, I assure you they are quite easy to learn and understand. I will make a detailed PR for the documentation regarding the 'Alert' feature so that you can gain an understanding of how it works and how it needs to be configured.

There are a few highly-rated GitHub repositories that can be used to implement the 'Alert' feature. Two of the most popular are python-telegram-bot and telethon. In this PR, I have used python-telegram-bot.

I have an Nvidia 3090 and use Windows, and I also have WSL installed. I would be more than happy to assist in testing the new features so that they can be released to the general public.