_bbox_transform_inv: Could not broadcast input array

[lunzueta]

Hi @mitmul. I'm testing this program in Windows 10 64-bits (Anaconda + Pycharm), following your instructions.

The first error I found was in the file cpu_nms.pyx, line 26: cdef np.ndarray[np.int_t, ndim=1] order = scores.argsort()[::-1], where the error was ValueError: Buffer dtype mismatch, expected 'int_t' but got 'long long'. This can be fixed by replacing int_t with int64_t (in a similar way as suggested in rbgirshick/py-faster-rcnn#36).

After fixing that error, I get another error in the file bbox_transform.py, line 73, in _bbox_transform_inv:
pred_boxes[:, 0::4] = pred_ctr_x - 0.5 * pred_w
ValueError: could not broadcast input array from shape (300,21) into shape (1,21)

I'm stuck with this. Any suggestion?

[mitmul]

Hi @lunzueta, I didn't notice that error because I haven't tested anything with Windows :p, so thank you for letting me know the first error. I added a description about the workaround you showed me to README.

As for the second error, that's a known bug of Chainer. Please see this PR to Chainer (and actually I wrote about this in my README.md, but sorry I guess it's hard to notice). I think some people have installed Chainer via pip, so it could be pain to apply diff to Chainer's source. So I've fixed this problem by replacing the Chainer's roi_pooling_2d function with my modified one at this commit.

Could you run git pull and try it again?

[lunzueta]

Thank you @mitmul. That solved the problem. Yes, I installed chainer via pip and I didn't realize about that comment in the description :-P