in the paper, the report run time for forward passing is <50ms in TiTan X gpu, however, I run the test demo for evaluating VOC2007 test set with python scripts provided by this repository, and used the Tesla K80 gpu, the running time it reports longer time. for example,
"im_detect: 4949/4952 net 0.140s preproc 0.016s postproc 0.001s misc 0.006s"
...
I do not change any other parameters, and I use the following command to run my test:
"./tools/test_net.py --net data/pvanet_models/PVA9.1_ImgNet_COCO_VOC0712plus_compressed.caffemodel --def models/pvanet/pva9.1/faster_rcnn_train_test_ft_rcnn_only_plus_comp.pt --cfg models/pvanet/cfgs/submit_1019.yml --gpu 0 --imdb voc_2007_test"
Can you provide me some more details for reproducing the results of the paper? I'm l very thanks for you help
in the paper, the report run time for forward passing is <50ms in TiTan X gpu, however, I run the test demo for evaluating VOC2007 test set with python scripts provided by this repository, and used the Tesla K80 gpu, the running time it reports longer time. for example, "im_detect: 4949/4952 net 0.140s preproc 0.016s postproc 0.001s misc 0.006s" ... I do not change any other parameters, and I use the following command to run my test: "./tools/test_net.py --net data/pvanet_models/PVA9.1_ImgNet_COCO_VOC0712plus_compressed.caffemodel --def models/pvanet/pva9.1/faster_rcnn_train_test_ft_rcnn_only_plus_comp.pt --cfg models/pvanet/cfgs/submit_1019.yml --gpu 0 --imdb voc_2007_test" Can you provide me some more details for reproducing the results of the paper? I'm l very thanks for you help