Faster-RCNN_TF icon indicating copy to clipboard operation
Faster-RCNN_TF copied to clipboard

How to fine tune faster-rcnn not from scratch?

Open yanmenglu opened this issue 7 years ago • 1 comments

hi @smallcorgi ,Thank you for your project.Now I have a problem ,could you help me? I need fine tune four classes(cat/dog/car/person),and this four classes are all in pascal_voc dataset.so I think if I can fine tune the weights that have pretrained from pascal-voc,but not from scratch(from VGG_imagenet.npy). I don't know am I right? if that right,could you tell me how to train from pascal-voc pretrained model from your project? Thank you very much.

yanmenglu avatar Sep 27 '17 01:09 yanmenglu

do u solve ur problem? i want do the same thing what u do

randomtutu avatar Jul 27 '18 06:07 randomtutu