Shivanand Venkanna Sheshappanavar

@dogyoonlee and @ashlet It seems like they are not doing rotation and jitter but only doing random point drop out, scaling, and shifting.

@SalaheddineSTA were you able to prepare your data?

@SalaheddineSTA Thanks for the update. I have a modified modelnet40 dataset with normals, x,y,z,nx,ny,nz data in .txt format (like the resampled data provided by the authors) and I am trying...

1) Set FLAGS.normal = True in train files. 2) Replace l0_xyz = point_cloud l0_points = None by l0_xyz = tf.slice(point_cloud, [0,0,0], [-1,-1,3]) l0_points = tf.slice(point_cloud, [0,0,3], [-1,-1,3]) in model files....

Seems like you are using the pointnet model instead of pointnet++ Please check if you want to raise this issue in pointnet github link.

Use below commands and see if it works for you. /usr/local/cuda/bin/nvcc --gpu-architecture=sm_35 --compiler-options -Wall -I/usr/local/cuda/include -L/usr/local/cuda/lib tf_sampling_g.cu -o tf_sampling_g.cu.o -c -O2 -DGOOGLE_CUDA=1 -x cu -Xcompiler -fPIC -lcusolver -lcurand -lcublas -lcusparse...

1) Set FLAGS.normal = True in train files. 2) Replace l0_xyz = point_cloud l0_points = None by l0_xyz = tf.slice(point_cloud, [0,0,0], [-1,-1,3]) l0_points = tf.slice(point_cloud, [0,0,3], [-1,-1,3]) in model files.

1) Set FLAGS.normal = True in train files. 2) Replace l0_xyz = point_cloud l0_points = None by l0_xyz = tf.slice(point_cloud, [0,0,0], [-1,-1,3]) l0_points = tf.slice(point_cloud, [0,0,3], [-1,-1,3]) in model files.

https://github.com/charlesq34/pointnet2/issues/41

I could reproduce only Trial1: eval mean mIoU: 0.816804 eval mean mIoU (all shapes): 0.847668 Trial2: eval mean mIoU: 0.814602 eval mean mIoU (all shapes): 0.848937 Trial3: eval mean mIoU:...