search

sfzhang15 / RefineDet

Single-Shot Refinement Neural Network for Object Detection, CVPR, 2018
Other
1.43k stars 393 forks source link

training own datasets, teset phase: out of memory #134

Closed gllspeed closed 5 years ago

gllspeed commented 5 years ago

Hi: I executed the script VGG16_VOC2007_512.py to train my own data set. The test phase of the training process reported error == cudaSuccess out of memory. The itersize was 1 in test.prototxt and the number of test pictures was 82. There is no problem with executing the script VGG16_VOC2007_320.py. What is the reason, thank you!

sfzhang15 commented 5 years ago

@gllspeed Batch_size=82 may be too large for the 512x512 input size, you can try to use small batch size.