07/12/24
Last week, recap:
merging image based funnel visits and proximity based tracking of funnel visits
Plan: use large radius around funnel for using proximity to detect when rat is nearby. Use image to detect when there ar actual visits. Use a conservative threshold to detect "nose pokes" and then a more liberal threshold to detect "drinking" (i.e. detect the difference between the rat checking the funnel vs running past (e.g., underneath) )
Progress:
remove redundancies from frame analyzer from image based method. Update to allow analysis of arbitrary numbers of dispensers (so that we can eventually analyze the dummy ones as well as the targets).
Updated code for behavioral scoring to use the more lenient proximity threshold and then to check the image based method for only epochs of close proximity for nose-pokes. Does not yet record 'dispenser pass-bys' where rat comes near dispenser but does not nose poke.
Plan for next week:
Reduce ambition of current scripts - annotate the videos with each of the parameters (disp proxy, nose pokes etc) for each frame
Create a struct with fields for each parameter (inc. notes for each regarding parameters / methods as approp.)
Can then define epochs later - allowing for epoch definitions to be changed easily and different sets of thresholds to be used and saved