Robin Cole
Robin Cole
This is probably a feature to leave until an official integration, since it would be a breaking change
Restrict to four point polygon - by combining multiple of these almost any desired shape can be created
Thanks @BrunoBarroso ! This is a good FR :) Already in https://github.com/robmarkcole/HASS-Deepstack-object - confidence: (Optional) The confidence (in %) above which detected targets are counted in the sensor state. Default...
@Minglarn events are fired with all info for each object, from these you can craft any behaviour you desire. Can you refine the feature request?
Each object is given a 'confidence', but for a given object this might change between captures (e.g. due to changing lighting) so cannot be relied on to identify an object....
@Minglarn thanks for putting thought into this. What would really assist the discussion is a demonstration of this algorithm on real data. I propose extracting a sequence of say 10...
You can use the https://www.home-assistant.io/integrations/folder_watcher/
Might be worth doing. In this image a bin is tagged as a car, and is similar size to the cars in the distance which appear small. 
I am waiting to see what, if any changes HA make to the implementation of image_processing. This integration is already complicated so not adding any new features until I know...
This could be achieved with an automation on events - what have you tried already? Adding "last_detection" could confuse people