Kongsea

I want a x86 Notepad2 copy supported MATLAB too. Can you send me a copy please? my email is konguo#gmail.com (please replace # with @)

I think you can build a version not think of MATLAB Stream comment. Because stream comment is not common.

More types of loss supported, such as `center loss`, `amsoftmax loss`, and so on.

@tweedmorris It seems your MD5 sums are the same with mine. So I think there is no problem with the files.

I also encounter this problem. After 29683 iters, it gives warnings: > train/../libs/boxes/bbox_transform.py:62: RuntimeWarning: overflow encountered in exp > pred_h = np.exp(dh) * heights[:, np.newaxis] > train/../libs/boxes/bbox_transform.py:61: RuntimeWarning: overflow encountered...

After 29683 iters, it gives warnings: > train/../libs/boxes/bbox_transform.py:62: RuntimeWarning: overflow encountered in exp > pred_h = np.exp(dh) * heights[:, np.newaxis] > train/../libs/boxes/bbox_transform.py:61: RuntimeWarning: overflow encountered in exp > pred_w =...

You need use python3 instead of python2.

exactly the same error as you. Is there anyone help to fix it?

Who can help to fix it?

Add `.astype(np.int)` to the original code in `lib/roi_data_layer/minibatch.py` to change it to: `fg_rois_per_this_image = np.minimum(fg_rois_per_image, fg_inds.size).astype(np.int)`