Update TOP 5 Use CasesToday, we want to give you a quick update on our customers’ TOP 5 USE CASES:
-> Detecting foxes & birds of prey around mobile chicken coops
-> Tracking coyotes in sensitive areas
-> Identifying drones for aerial security
-> Smart object detection for logistics operations
-> Differentiating your dog from other animals in your garden
👉 Which challenge can we help you solve?
#cctv #security #surveillance #smarthome #ai
02.02.2025: v1.6.6CAM-AI-Proxy: Switching alarms far away from your server
- We added a plugin-interface for our new CAM-AI-Proxy to the CAM-AI server. The Proxy is a software running on a Raspberry Pie distant from the server that can perform alarms (for example switching a siren or a blinking light). Thus the alarm devices (and the cameras, too) can be placed anywhere on the internet. For details ask us: ...*protected emai
- Some minor bug fixings & improvements.
28.01.2025: v1.6.5Tag name editor & searching for cameras on remote subnets
- If you want to give your image categories different names, because you are training on non standard image content items, you can do so now easily. You will find the relevant masks under Main --> Tools --> Configuration --> Tags.
- When connecting new cameras in expert mode, you can now scan for remote cameras on remote ip ranges using a subnet filter. This is very handy when connecting cameras via the new CAM-AI-Proxy. You can easily search the addresses that the proxy opened for you.
23.01.2025: v1.6.4Some minor bug fixings
The most important ones:
- Until now, it was possible, to launch a training from a training server. That caused problems because of the incomplete data situation. This option is now switched of.
- When having launched a training, the "Train Now" - button would be enabled for a couple of seconds after the launch. That was no good, because it resulted in the user being able to launch the same training two times. No good, but now fixed.
20.01.2025: v1.6.3Eventer watchdog / simultaneous training / bug fixing
We have three not very visible but important improvements:
- Under certain conditions runtime errors in the eventer would make the eventer stop working, resulting in no more alarms, emails etc. Now the watchdog will detect such a situation, shut down the eventer and relaunch it. The errors are still noted in the logfile.
- There are now more than one training process on the training servers available, provided that the GPU has enough RAM. In our training server in Eygelshoven we have enough space for 5 training jobs at hte same time. That should be enough to avoid queues in front of the trainer. We recommend to upgrade your local server (raspi etc.) also, because for taking profit from this, you need to have version 1.6.3 on your raspi. Otherwise you will only have trainer process number 1 available.
- A bug in training was fixed: Under some conditions the trainer did not take in all new images to the training, sometimes resulting in poor adaptation to the last added images. This is fixed now.
13.01.2025: v1.6.2SMTP, configuration, improvements for two user screens
We recommend to upgrade all servers
We completely redid SMTP (email sending). TLS is now set automatically if supported by the server. The SMTP-client is now much more reliable than before. Errors during sending of emails are noted and described in the logfile.
The new configuration menu still has only one working submenu: SMTP. Although this is now based on our new SMTP, there is not much difference on the surface: The option "use_tls" is gone and the error messages for the guided trial-and-success of SMTP setup are now more helpful.
Two user screens were renovated behind the surface:
events ---> [camera name]Â and schools ---> [school name]
These two should now be much easier to use (especially on Raspis), because the prediction lists for the display are now being prepared in advance in the background, enabling smoother scrolling.
The user screen preferences ---> [stream name] (with the green and red frames) performed some confusing visualizations. Now this is streamlined: Under "Conditions for showing event frames" you select, which events you want to see. The color of the frames is set depending on the condition set (school, video, email, alarm) you opend last:
Condition met > red frame, condition not met > green frame
28.12.2024: v1.6.1Configuration & Logfile sending
We improved the "Send Logfile" feature further and we introduced a new "Configuration" menu. The SMTP setup is only the start, more configuration submenues are going to come...
02.12.2034: v1.5.9Logfile sending option and Debugging
- There is a new feature: Wenn the CAM-AI support asks you to send the logfiles for problem analysis, you can from now on do so with just two mouseclicks (see screenshot). Please keep in mind: To use this feature you need to have correct SMTP-settings (in passwords.py).
- There was a bug in the timing for the detector that cause hangers, especially on Raspis. Shpuld be fixed.
- Some minor bugfixes and some cleanup...
28.11.2024: v1.5.7Improvements and bugfixes
- The new virtual camera still had some timing issues. No its stabilized. When operated with no fps limit, the video files is processed at maximum possible speed, not the recording speed. It is good to have a fast computer for this kind of work.
- The eventer preview ran bumby in some situations. THis is now fixed.
- Sending Emails did not work with lots of large images attached. Now it does.
- There was a problem in the trainer. Some jobs got stuck. Now they all go.
24.11.2024: v1.5.5Virtual Camera & Bugfixes
- We have a new feature, the "Virtual Camera". This enables training on Video Files, for example footage from your camera from the past. Very quick user instructions:
a) Install your virtual camera from the MainMenu>Cameras>Virtual camera installation by selecting the file from your Computer. After uploading to the server, the video will run in an endless loop and the Motion Detection, Event Generation etc. will be performed based on the video data from that until you uninstall the Virtual Camera. Please use this feature only if you know what you are doing...
- The server shutdown from main menue did not work. Now it does.
- There was a problem in the Eventer Preview. Should work much better now.
- A couple of minor bugfixes...
19.11.2024: v1.5.1Bugfixes (Upgrade recommended for Raspi Users):
- In release 1.5.0 there was a bug that sometimes prevented the Database from being updated correctly. We fixed that one.
- The Grid view is now working in all three places: Cameras, Detectors and Preferences
17.11.2024: v1.5.0 (Part 2)Abstract: Grid view, LiteRT and Conda plus a lot more
- We have activated the grid view that will display your cameras in the most space saving way. You will find this option in the submenues for Cameras, Detectors and Preferences. Unfortunate the preferences grid is not yet working. It will in the next release.
- We now have a quota system sending warning emails if the HD or SD space gets tight and prevents from filling up the space completely.
- We can now configure distributed systems sharing the workload. For example, we used that for our two servers, one (in Karlsruhe) doing the web service (django.cam-ai.eu) and one (in Eygelshoven) doing the training.
- The inferencing on the Raspi servers is now done with LiteRT-Models that are specifically designed for small servers.
- The Python environment for the server process on the Raspi is now Conda. The same as it is on the PC installations.
- CAM-AI was adapted to run with the most recent versions of Python, Django, Channels and Tensorflow.
- ...lots of little improvements and bug fixings...
17.11.2024: v1.5.0 (Part 1)...differently from the previous versions, the Raspi server is now based on a Conda environment for the Python libraries (like the PCs are). This change requires the installation of Conda before you perform the server software upgrade from the CAM-AI main menu. Here is how. After logging into the Raspi you have to perform these Steps:
sudo systemctl disable dphys-swapfile
sudo systemctl stop dphys-swapfile
wget https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-Linux-aarch64.sh
bash Miniforge3-Linux-aarch64.sh
This installation script is a little odd. First you press Enter to start installation. After reading the license agreement you hit Ctrl-C and Enter again. After that you confirm all other questions with either yes or Enter for default (the installation directory).
rm Miniforge3-Linux-aarch64.sh
source ~/.bashrc
conda create --name tf python=3.11
If your CAM-AI server is configured for autostart, you have to change the autostart script slightly:
nano ~/bashes/start-c_server-service.sh
Find the line
source env/bin/activate
and replace it with these two lines:
source ~/miniforge3/etc/profile.d/conda.sh
conda activate tf
When you are done, save and reboot your Raspi.
sudo reboot now
After that you should be able to upgrade as usual...
14. 09. 2024: v1.4.7Abstract: We improved speed and reliability for remote training. Please update if you have problems with training.
– In the past the remote training (done by our servers in Eygelshoven and/or Karlsruhe) was slow and not always 100% stable. We have changed the protocol of this process. Now its well designed and fast.
12. 09. 2024: v1.4.6Abstract: Please update as soon as possible. We found and fixed an instability.
- We found an instability with client systems running on fast hardware. There was a timing issue in the protocol for clients (Raspis etc.) requesting predictions (= inference) from the server. Therefore we had to change the protocol on both sides. After updating the server the clients might not work anymore.
