What is the correct syntaxt in config.yml to apply a mask to an object?
e.g. I want to mask a parking place only for object car and not for persons etc…
Their is mask maker in Frigate UI>>cameras>>select camera live view>>show options>>mask and zone creator
You can create mask/zone and then copy/paste into your config. Just correct spacing
Sorry that is not an answer to my question. This part I already had… I asked how to get the mask in for a typical type of object and then ask for an example with correct spacing.
example for “person” in documentation for “objects”
Yes, I have seen that but there is no information I can find for the filter: part
(see my other question Local realtime person detection for RTSP cameras - #4686 by sender)
And this example is not clear for me regarding this part of my question:
So if this is my camera config:
schuurachter:
ffmpeg:
inputs:
- path: rtsp://admin:[email protected]:554/Streaming/Channels/101/
roles:
- detect
- clips
# - rtmp
width: 2048
height: 1536
fps: 4
snapshots:
enabled: true
crop: false
bounding_box: True
objects:
track:
- person
- car
- dog
- bicycle
- cat
motion:
mask:
- '443,106,775,108,679,365,294,371'
- '2048,371,2048,585,1136,401,1090,130'
clips:
# Required: enables clips for the camera (default: shown below)
# This value can be set via MQTT and will be updated in startup based on retained value
enabled: True
# Optional: Number of seconds before the event to include in the clips (default: shown below)
pre_capture: 2
# Optional: Number of seconds after the event to include in the clips (default: shown below)
post_capture: 2
# Optional: Objects to save clips for. (default: all tracked objects)
objects:
- person
- car
- dog
- bicycle
- cat
# Optional: Restrict clips to objects that entered any of the listed zones (default: no required zones)
required_zones: []
# Optional: Camera override for retention settings (default: global values)
retain:
# Required: Default retention days (default: shown below)
default: 10
# Optional: Per object retention days
objects:
person: 100
car: 50
dog: 50
cat: 10
bicycle: 50
how would that look like to have cars not detected in the masked zone but do have person, dog, bicycle and cat in it?
filter is where object specific config parameter are placed
In this case I don’t think there is further detail to be provided since those items are documented in camera level documentation
objects:
track:
- person
- car
# Optional: mask to prevent all object types from being detected in certain areas (default: no mask)
# Checks based on the bottom center of the bounding box of the object.
# NOTE: This mask is COMBINED with the object type specific mask below
mask: 0,0,1000,0,1000,200,0,200
filters:
car:
min_area: 5000
max_area: 100000
min_score: 0.5
threshold: 0.7
# Optional: mask to prevent this object type from being detected in certain areas (default: no mask)
# Checks based on the bottom center of the bounding box of the object
mask: 0,0,1000,0,1000,200,0,200
I do not get that documentation. I can’t get it working because after reading it I do not know where in what order and indention to place. How do I split the
objects:
track:
- person
- car
- dog
- bicycle
- cat
to only have car match the mask?
A mask prevents either motion, or an object, being detected in a masked area. It is not a ‘match only this in the masked area’ mask.
There is no ‘inverted mask’, as far as I’m aware, you’d have to mask all other objects, all around the area you don’t want them to be detected in.
So, to NOT match a person in 95% of the image, you’d have to mask person in 95% of the image.
I understand. But what do I practically need to do to:
to prevent detection of cars in a parking space
to not prevent and other object detection on that parking space
Currently on every “motion”, wind, light, clouds I have lists full of cars that are parked.
And please help me with the yaml layout.
To prevent the car being detected in the parking space, create a car object mask so it covers the bottom-middle of the bounding box. Detection is based on the bottom middle point of the bounding box, so you need to mask this.
In my case, I mask about the bottom 10% of my garage, with a car mask… this means it can be detected as it approaches the garage, but not when it is parked (because when it is parked, the bottom middle of the bounding box is in the bottom 10% of the image)
Sorry I dont get that… do you have a picture of it?
You’re welcome to post pictures of you car, garage (with the bounding box), etc on a public forum and I’ll suggest where I’d mask - but I’m not going to be posting mine 
Just a random google image:
this is where I would not like to detect cars but do want to have people, dogs, etc if there are no cars:
Not useful - to mask, you need to know where it is being detected. And that is the bounding box. Mask the bottom centre of wherever it is being detected, to prevent it being detected, without masking the entire area.
If red is the bounding box, green is where you could mask to prevent it being detected - because it covers the bottom centre/middle of the box.
Pretty sure @tmjpugh already answered that in post 4692:
Where ‘mask: 0,0,1000,0,1000,200,0,20’ is your black area - that would prevent cars only being detected in that area.
objects:
track:
- person
- car
- dog
- cat
- bicycle
filters:
car:
min_area: 5000
max_area: 100000
min_score: 0.5
threshold: 0.7
mask: 0,0,1000,0,1000,200,0,200
1 Like
Anyone using frigate with nginx proxy?
how you do this? (config example)
So I did this and Frigate restarted ok, will let you know if that works (not listing cars anymore on the parking.
schuurachter:
ffmpeg:
inputs:
- path: rtsp://admin:[email protected]:554/Streaming/Channels/101/
roles:
- detect
- clips
# - rtmp
width: 2048
height: 1536
fps: 4
snapshots:
enabled: true
crop: false
bounding_box: True
objects:
track:
- person
- car
- dog
- bicycle
- cat
filters:
car:
min_area: 5000
max_area: 100000
min_score: 0.5
threshold: 0.7
mask:
- '443,106,775,108,679,365,294,371'
- '2048,371,2048,585,1136,401,1090,130'
clips:
# Required: enables clips for the camera (default: shown below)
# This value can be set via MQTT and will be updated in startup based on retained value
enabled: True
# Optional: Number of seconds before the event to include in the clips (default: shown below)
pre_capture: 2
# Optional: Number of seconds after the event to include in the clips (default: shown below)
post_capture: 2
# Optional: Objects to save clips for. (default: all tracked objects)
objects:
- person
- car
- dog
- bicycle
- cat
# Optional: Restrict clips to objects that entered any of the listed zones (default: no required zones)
required_zones: []
# Optional: Camera override for retention settings (default: global values)
retain:
# Required: Default retention days (default: shown below)
default: 10
# Optional: Per object retention days
objects:
person: 100
car: 50
dog: 50
cat: 10
bicycle: 50
what are these for:
I cannot find an explanation in the documentation. only this https://blakeblackshear.github.io/frigate/configuration/false_positives/ but that does not explain met the meaning and values of min_area and max_area.
1 Like
I highly recommend spending some time reading the documentation start to finish, multiple times, including the commented/hashed lines in the example configuration - just reading, without trying to configure anything, or without skipping back and forth.
min_area and max_area are the area sizes, they’re shown on the bounding box on the snapshots if enabled.
Objects, bottom of the above:
# Optional: minimum width*height of the bounding box for the detected object (default: 0)
# Optional: maximum width*height of the bounding box for the detected object (default: 24000000)
… over and out from me, I need a drink
P.S. You might actually need to at least remove the min_area and/or max_area, depending on the size of the cars being detecting - the numbers on the bounding box snapshots, if enabled, will tell you what the sizes are.
You need to look at a few of the false positives, to work out what the range of sizes are, for example, my car as it enters through the gates is around 10000 (so this is min_area, roughly, if i want it to be detected at that distance)… as it enters the garage and comes closer to the camera it is around 40000, but when parked and as close to the camera as it is going to be it is around 100000.
Naturally the above varies depending on the size of the car, how close the car gets to the camera and/or how far away it can reliably be detected.
Can’t I then better just leave these out?