d111cdae69
Updates to all parts of model building - moving to frozen transfer learning followed by slowed learning rate fine tuning using EfficientNets for final model.
Lucas2022-06-03 13:44:34 -04:00
755fcde3a9
clean up imports. fix naming, force CPU to fill the cache faster with images using 20 workers.
Lucas2022-06-01 18:52:55 -04:00
1b539d6945
renaming all files - moving training to be a single file for transfer vs not transfer learning. Made the testing file test all models. Needs to be updated to only update with new models.
Lucas2022-06-01 17:46:55 -04:00
616c99c7f5
creating model builder for faster model tweaking and iteration. Increased threads for training to better feed the GPU images from the image generator.
Lucas2022-05-25 19:47:30 -04:00
5990351a47
moving to Tensorflow 2.0 and changing the test train split to use symlinks.
Lucas2022-05-18 18:37:29 -04:00
24e8d71306
Made all processes threaded. Also changed up the neural network to hopefully better support the higher class load
Lucas Oskorep2019-04-26 02:57:18 -05:00
ef71399199
Updating the input image sizes to have a higher resolution. Currently unclean data is reaching 70% accuracy.
Lucas Oskorep2019-04-14 18:55:28 -05:00
fac82296e2
Updating the input image sizes to have a higher resolution. Currently unclean data is reaching 70% accuracy.
Lucas Oskorep2019-04-14 18:54:26 -05:00
de117a55ae
Updating the input image sizes to have a higher resolution. Currently unclean data is reaching 70% accuracy.
Lucas Oskorep2019-04-14 18:50:53 -05:00
63c6ae62b3
Updating the input image sizes to have a higher resolution. Currently unclean data is reaching 70% accuracy.
Lucas Oskorep2019-04-14 15:19:23 -05:00