search

EdjeElectronics / TensorFlow-Object-Detection-API-Tutorial-Train-Multiple-Objects-Windows-10

How to train a TensorFlow Object Detection Classifier for multiple object detection on Windows
Apache License 2.0
2.93k stars 1.3k forks source link

out of memory when training #79

Open ghost opened 6 years ago

ghost commented 6 years ago

when I am training , the cup occupied memory at 93% even more. the model I used is faster rcnn inception resnet v2 and the size of TF_record is 117G. I was wondering that if it was due to the data was too huge? so. how can I handle this problem? sorry for my bad English and thank you for you attention.

leccyril commented 6 years ago

you maybe have too big picture to train, you can resize it in model configuration file, i read it add in queue all picture size...

leccyril commented 6 years ago

CUP or CPU ? cpu it is normal because it use more CPU than memory, memory is used for keep picture in memory... but OOM problem while training is a picture size issue

ghost commented 6 years ago

@leccyril thank you for your professional answers. my picture size is 1069*500，and indeed it shows 00M.

leccyril commented 6 years ago

Tf record is to big try with 300x300