Attendees: D.Cline, D.Edgington, J.Ellena, L.Kuhnz, K.Smith, M.Vadaro

Danelle completed work on AVED still frame modifications and processed some test StaM image sets. Thanks for Linda for help on improving the detection results and understanding the animals of interest !

Results were discussed and it was decided to AVED process a 400 foot roll of M.Vardaro's choosing and to look for Echinocrepis rostrata and create a training database from it. This 400 ft. roll will be approximately 3200 individual still images taken in 1 hour time-lapse increments.

Lots of insightful questions were discussed about the AVED results from the Sta4211 dive, including

• why certain objects were not bounded correctly
• how to deal with the timecode overlay (which skews the results because it's salient)
• how the AVED classifier voting works
• how to use size/color/time to tell whether an animal is the same individual or not
• using Photoshop batch processing for some of the necessary pre-processing steps

Next meeting, the functional requirements document will be reviewed in detail.

Linda will be working some test images using Photoshop to erase the timecode overlay for Danelle to test.

Danelle will be working on engineering the workflow to process the data set.