Closed andrewhuang121 closed 4 years ago
Update: GPU instance now has slow-but-HQ pipeline working.
Was able to shave off appx 10 minutes off computation time by overriding certain feature extraction parameters
Built fast pipeline... appx 1hr faster but still takes a while. Will try another speed/quality test with smaller object more similar to what we wanted to do.
TODO: -modify generation of comp. graphs -time difference -look at quality difference -maybe convert pipeline into a bash file to prevent python related memleaks