search

fizyr / keras-retinanet

Keras implementation of RetinaNet object detection.
Apache License 2.0
4.38k stars 1.96k forks source link

Where does train.py save and can it be stopped early? #1389

Closed nsgalea closed 4 years ago

nsgalea commented 4 years ago

Hi I am just getting started with ML and am running the train.py on a custom dataset as a test to try and figure out how things work. It's been pretty difficult finding good tutorials on training models, since most of the ones I tried didn't work. I finally managed to get train.py running on my own dataset using the guide in the readme, but I am still not 100% sure how it works.

It says it is going to run for about 80 hours more. I am wondering if it is possible to stop early with maybe a less accurately trained model? Also does it actually save somewhere on my HD when it is done? I was looking at the code and couldn't figure out if/how it saves.

Sorry for asking here, I don't know if this is the best place since it isn't really an "issue" with keras-retinanet, so much as an issue with my understanding of how it works.

Gateway2745 commented 4 years ago

You can always stop your model after training for few epochs .The model will be saved in ./snapshots by default after every epoch.

stale[bot] commented 4 years ago

This issue has been automatically marked as stale due to the lack of recent activity. It will be closed if no further activity occurs. Thank you for your contributions.