Post processing video magic with ZoneMinder: find missing objects, blend multiple events, annotate videos, and more.
zmMagik will be a list of growing foo-magic things you can do with video images that ZM stores. Probably...
As of today, it lets you:
- Blend multiple events to quickly see how the day went. Imagine compressing 24 hours of video into 1 minute with object overlays. Gadzooks!
this video is blended from 2 days worth of video. Generated using
python3 ./magik.py -c config.ini --monitors 11 --blend --display --download=False --from "2 days ago"
- Annotate recorded ZoneMinder videos. Holy Batman!
python3 ./magik.py -c config.ini --eventid 44063 --dumpjson --annotate --display --download=False --onlyrelevant=False --skipframes=1
- Find an image fragment inside multiple events. For example, someone stole your amazon package. Crop a picture of an event with that package and then ask zmMagik to search for events where this package went missing. Great Krypton!
python3 ./magik.py -c config.ini --find trash.jpg --dumpjson --display --download=False --from "8am" --to "3pm" --monitors 11
I came home one day to see my trash can cover went missing. I thought it would be fun to write a tool that could search through my events to let me know when it went missing. Yep, it started with trash talking
Andy posted an example of how other vendors blend multiple videos to give a common view quickly. I thought it would be fun to try
One thing leads to another and I keep doing new things to learn new things..
- Only works with video mp4 files. Did not bother adding support for JPEG store
- Very Beta. Also, if you don't have a GPU, make sure you play with the flags to optimize skipframes, detection mode, resize
- Multi-server most likely won't work
# needs python3, so you may need to use pip3 if you have 2.x as well git clone https://github.com/pliablepixels/zmMagik cd zmMagik # you may need to do sudo -H pip3 instead for below, if you get permission errors pip3 install -r requirements.txt
If you are using yolo extraction, you also need these files and make sure your config variables point to them
wget https://raw.githubusercontent.com/pjreddie/darknet/master/cfg/yolov3.cfg wget https://pjreddie.com/media/files/yolov3.weights wget https://raw.githubusercontent.com/pjreddie/darknet/master/data/coco.names
General note: do a
python3 ./zmMagik -h to see all options. Remember, you can stuff in regularly used options in a config file and override on CLI as you need. Much easier.
- Create a blended event for monitor 11 in specified time range. config.ini has your ZM usename/password/portal etc so you don't need to type it in every time
python3 ./magik.py --monitors=11 --from "yesterday, 7am" --to "today, 10am" --blend -c config.ini
- Search when an object (image) went missing:
python3 ./magik.py --monitors=7 --present=False --from "today, 7am" --to "today, 7pm" --find "amazonpackage.jpg" -c config.ini
amazonpackage.jpg needs to be the same dimensions/orientation as in the events it will look into. Best way to do this is to load up ZM, find the package, crop it and use it.
How do I use GPU acceleration?
- See GPU section below
What is "mixed" background extraction?
- This is the default mode. It uses the very fast openCV background subtraction to detect motion, and then uses YOLO to refine the search to see if it really is an object worth marking. Use this mode by default, unless you need more speed, in which case, use "backround_extraction"
Using "background_extraction" mode isn't that great
- Yes, that's why you should use "mixed"
- Some tips:
- Use masks to restrict area
--displayto see what is going on, look at frame masks and output
- Try changing the learning rate of the background extractor
- See if using a different Background extractor for
globals.pyhelps you (read this)
- Fiddle with kernel_clean and kernel_fill in
Using "Yolo" or "mixed" extraction mode is great, but it overlays complete rectangles
- Yes, unlike "background_extraction" yolo doesn't report a mask of the object shape, only a bounding box
- I'll add masked R-CNN too, you can try that (will be slower than Yolo)
- Maybe you can suggest a smarter way to overlay the rectangle using some fancy operators that will act like its blending?
finddoesn't find my image
- Congratulations, maybe no one stole your amazon package
- Make sure image you are looking for is not rotated/resized/etc. needs to be original dimensions
As of today, OpenCV DNN doesn't support GPU acceleration for NVIDA (no idea about non NVIDIA). This may change soon with a work in progress contribution to get it working with CUDA. Till that happens, we need to directly use a GPU enabled version for YoloV3. I went with the darknet fork maintained by AlexyAB
Compile it with GPU
Only builds of darknet from 2019-10-25 or later should be used with zmMagik due to a change in the darknet data structures
Make sure it is actually using GPU
darknet_lib=<path/to/filename of gpu accelerated so>
If you need help compiling darknet for GPU and CUDA 10.x, see simpleYolo
Do NOT use darknet lib directly for a CPU compiled library. It is terribly slow (in my tests, OpenCV was around 50x faster)
How much GPU memory do I need?
- The YoloV3 model config I use takes up 1.6GB of GPU memory
- Note that I use a reduced footprint yolo config. I have 4GB of GPU memory, so the default
yolov3.cfgdid not work and ate up all my memory. This is my modified
yolov3.cfgsection to make it work:
[net] batch=1 subdivisions=1 width=416 height=416 <and then all the stuff that follows>
How much speed boost can I expect with GPU?
- Here is a practical comparison. I ran a blend operation on my driveway camera (modect) for a full day's worth of alarmed events. I used 'mixed' mode, which first used openCV background subtraction and then YOLO if the first mode found anything. This was to be fair to the CPU stats when compared. It grabbed a total of 27 video events:
python3 ./magik.py --blend --from "1 day ago" --monitors 8 -c ./config.ini --gpu=True --alarmonly=True --skipframes=1 Total time: 250.72s
I then ran it without GPU: (Note that I have
libopenblas-dev liblapack-dev libblas-devconfigured with OpenCV to improve CPU performance a lot)
python3 ./magik.py --blend --from "1 day ago" --monitors 8 -c ./config.ini --gpu=False --alarmonly=True --skipframes=1 Total time: 1234.77s
Thats a 5x improvement
- On my 1050 Ti, YoloV3 inferences drops to 120ms or less, compared to 2-3 seconds on GPU
- That being said, blending/annotating involves:
- reading frames (A)
- processing frames (B)
- writing frames (C)
- GPU affects point B. If you are reading very large events, A & C will still take its own time. You likely won't see a big improvement there. If there are many objects (B), then obviously, GPU performance improvements will have a huge impact. To make A & C faster:
- If you use mocord in ZM, then starting ZM 1.34, set
EVENT_CLOSE_MODEto "alarm". That will create a new event when an alarm occurs, and close it when the alarm closes. That will help you speed things up a lot
- All that being said, I'm using a threaded opencv pipeline to read frames which does improve read performance compared to before (credit to imutils)