jugg1024

Hi, any progress here? Could you please offer example codes of lintelmodule.c for python 2.x?

I've already put `libwarpctc.so` into `$LD_LIBRARY_PATH` but `setup.py build_ext --inplace` still can not find `-lwarpctc` by change `-lwarpctc` into `-L/my/path/to/libwarpctc.so` compilation done

@Bartzi thx for reply, its in my local path, not system path like `/usr/lib`.

I'm sorry to reply you so late, you should use ` ./experiments/scripts/faster_rcnn_end2end.sh 0 VGG16 pascal_voc` because training data are prepare in the format of pascal_voc although we use coco_text currently,...

hi @athmey , i also use the default settings of frcn, but the pretrain model i use is official VGG16. And do you use full coco-text training data?

Hi, i wrote it on readme.md， you should prepare your dataset(both data and label) in the format of pascal_voc annotation, examples can be seen in: ./datasets/script/format_annotation.py format_coco_text() then training script...

it seem you have some problems on build caffe-fast-rcnn, it tells that you should add option -fPIC. you can try it. and can u build the offical caffe in on...

yes, I use coco_text.

yes, fine-tuning vgg network using coco-text dataset, only use 'machine printed' and 'legible' text data, neg sample will be generate by frcn according to iou