search

unsky / FPN

Feature Pyramid Networks for Object Detection
524 stars 263 forks source link

Does the code support training fpn with batch size 1 with one gpu? #83

Open anorthman opened 6 years ago

anorthman commented 6 years ago

Does the code support training fpn with batch size 1 with one gpu?

unsky commented 6 years ago

yes

mantou22 commented 6 years ago

when I run the demo.py,there is a problem,could you help me??????

Loaded network /home/xmpu234/Documents/FPN-caffe_1/data/faster_rcnn_models/fpn_iter_25000.caffemodel Traceback (most recent call last): File "/home/xmpu234/Documents/FPN-caffe1/tools/demo.py", line 142, in , _= im_detect(net, im) File "/home/xmpu234/Documents/FPN-caffe_1/tools/../lib/fast_rcnn/test.py", line 158, in im_detect rois2 = net.blobs['rpn_rois/p2'].data.copy() KeyError: 'rpn_rois/p2

anorthman commented 6 years ago

@unsky ok

anorthman commented 6 years ago

@mantou22 no layer named "rpn_rois/p2"

mantou22 commented 6 years ago

@anorthman there is this layer named “rpn_rois/p2”

mantou22 commented 6 years ago

when I run the demo.py,there is a problem,could you help me?????? Traceback (most recent call last): File "/home/xmpu234/Documents/FPN1/tools/demo.py", line 144, in , _= im_detect(net, im) File "/home/xmpu234/Documents/FPN_1/tools/../lib/fast_rcnn/test.py", line 298, in im_detect box_deltas = box_deltas * stds + means ValueError: operands could not be broadcast together with shapes (1000,8) (84,)

mantou22 commented 6 years ago

作者您好. 训练集为100100左右图片， 目标大小为4040左右，请问如何设置anchor的大小

unsky commented 6 years ago

可以参考yolo对数据集进行分析 进行聚类

mantou22 commented 6 years ago

您好，请问为什么我训练自己的模型会怎么慢速度？ 前面几百次迭代，每20次只需要 34S 左右， 到了2000次左右时， 每20次需要 5-6分钟 为什么会那么慢？？？？？ 我的显卡是 GPU 1080Ti 的 12G运行内存