search

Sense-X / TSD

1st place models in Google OpenImage Detection Challenge 2019
Apache License 2.0
456 stars 64 forks source link

The x101_64x4d model of train and inference is very very slow? #6

Closed lexiqi closed 4 years ago

lexiqi commented 4 years ago

when i train faster_rcnn_x101_64x4d_fpn_TSD.py, the train and inference is very slow. The inference speed is just 0.8 task/s。 (the others model is normal)

songguanglu commented 4 years ago

Hi, limited by the GPU memory, the checkpoint mechanism is used when training faster_rcnn_x101_64x4d_fpn_TSD.py, and the multi-scale test will slow down the inference speed. You can set the with_cp=False with imgs_per_gpu=1 to perform training.

lexiqi commented 4 years ago

ok，get√，thank you!!

jiangyangzhou commented 3 years ago

Hello, I test faster_rcnn_x101_64x4d_fpn_TSD.py on 1 V100, and the inference speed is only 0.2 task/s, is it a normal phenomenon ? Thanks.