search

unsky / FPN

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

another problem #63

Open sujinpeng opened 6 years ago

sujinpeng commented 6 years ago

Traceback (most recent call last): File "./tools/train_net.py", line 112, in max_iters=args.max_iters) File "/media/E/FPN-caffe4/tools/../lib/fast_rcnn/train.py", line 160, in train_net model_paths = sw.train_model(max_iters) File "/media/E/FPN-caffe4/tools/../lib/fast_rcnn/train.py", line 101, in train_model self.solver.step(1) File "/media/E/FPN-caffe4/tools/../lib/rpn/anchor_target_layer.py", line 132, in forward if int(item)==p: ValueError: cannot convert float NaN to integer

unsky commented 6 years ago

numpy version is too low

sujinpeng commented 6 years ago

thanyou very much !! I habe a question,in your code k = np.floor(k0 + np.log2(np.sqrt(areas)*1.0 / 64)) could you tell me why is the 64? in the paper ,it's 224

unsky commented 6 years ago

for voc anchor size setting

sujinpeng commented 6 years ago

Thank you very much. Then,I found the result of the cfg.TEST.BBOX_REG=False is better than the cfg.TEST.BBOX_REG=True, why?

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

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