search

healthonrails / annolid

An annotation and instance segmentation-based multiple animal tracking and behavior analysis package.
Other
39 stars 9 forks source link

Glitter2 Annolid tracking results incompatibility #51

Open lukeee24 opened 9 months ago

lukeee24 commented 9 months ago

Hi, have trained an Annolid model and used batch inference to generate tracking .csv's for 30 videos I'd like to analyse animal position changes (3 zones) - I cannot seem to get glitter2 to recognise the tracked mice as a 'pos' at all though? Any suggestions? Using .csv import tool on glitter2 isn't working for the tracked.csv (missing first three columns needed by the programme) or the tracked.nix.csv - issue with framenumber column, which can't be fixed by removing this?

healthonrails commented 9 months ago

I'm sorry to learn about the issue. Have you made any changes to the Colab notebook? Additionally, could you kindly provide the initial rows of the tracked.csv file as an example?

lukeee24 commented 9 months ago

Thanks for the quick reply! No changes. Initial rows of tracked.csv: frame_number x1 y1 x2 y2 cx cy instance_name class_score segmentation tracking_id

Initial rows of tracked_nix.csvs produced: metadata value timestamps event:Grooming event:Rearing event:Object_investigation event:RightInteract event:LeftInteract event:Freezing pos:animal_center:x pos:animal_center:y

healthonrails commented 9 months ago

Can you capture screenshots and provide a detailed explanation of your intended outcome? Additionally, if you plan to showcase the subjects in the video, please input their names as a comma-separated list in the Glitter2 Dialog.