search

zhaoweicai / mscnn

Caffe implementation of our multi-scale object detection framework
404 stars 211 forks source link

questions about training on new dataset #91

Open misstong opened 6 years ago

misstong commented 6 years ago

how to generate window files from voc format,is there any tool available? does the size of input images have some limitation?

Hub-Tian commented 6 years ago

Using the code here (https://github.com/umautobots/vod-converter), I transformed the voc data to kitti format and generated the window files. But the training process is really tough for me and I can't get good results. Do you have any idea?