something about demo

[HIT-LWQ]

Dear shuran,
I'm very interested in the SSCNet. Recently, I try to run the demo,but there are some ploblems. (I test using cpu)
firstly, Check failed: num_axes()<=4(5 vs. 4) Cannot use legacy accessors on Blobs with >4 axes. So I commented out code "CHECK_LE(num_axes(), 4)
<< "Cannot use legacy accessors on Blobs with > 4 axes.";
CHECK_LT(index, 4);
CHECK_GE(index, -4); " in "sscnet/caffe_code/caffe3d_suncg/include/caffe/blob.hpp"
secondly, When I run the demotest_model.py, the result of classlabel.max() and min() in the code
"def vol2points(vol,tsdf,seg_label):
classlabel = np.argmax(vol, axis=1)
colorMap = np.array([[ 22,191,206],[214, 38, 40],[ 43,160, 43],[158,216,229],[114,158,206],[204,204, 91],[255,186,119],[147,102,188],[ 30,119,181],[188,188, 33],[255,127, 12],[196,175,214],[153,153,153]])
points = []
rgb = []
for x in range(classlabel.shape[1]):
for y in range(classlabel.shape[2]):
for z in range(classlabel.shape[3]):
tsdfvalue = tsdf[0][0][4x][4y][4*z];
if (classlabel[0][x][y][z] > 0 and seg_label[0][0][x][y][z] <= 254 and ( tsdfvalue < 0 or tsdfvalue > 0.8)):
points.append(np.array([x,y,z]))
rgb.append(np.array(colorMap[classlabel[0][x][y][z],:]))
points = np.vstack(points)
rgb = np.vstack(rgb)
return {'points':points, 'rgb':rgb}"
are all zero, it's because I using the cpu?
Thank you for your help.

[shurans]

@761687049 The current code doesn't support CPU mode, could you try to run it under GPU mode?

[HIT-LWQ]

let me try again, thanks

[LongruiDong]

let me try again, thanks

Hi,I am facing the same issue with your first problem

Check failed: num_axes()<=4(5 vs. 4) Cannot use legacy accessors on Blobs with >4 axes

I am using gpu but without Cudnn (Comment USE_CUDNN := 1 )，have you solved this problem?