Bo Yang

Hi @RuYi-Pan , please refer to the proposed Algorithm 1, "theta1" --> 100, "theta2" --> 20, the two lines you mentioned are the implementation of "step 2".

hi @RuYi-Pan: step 1: if the point P is within the pred bbox [B_min, B_max], delta_xyz will be positive, if P is outside, then negative. step 2: to clip the...

Hi @fxy2012 , I would suggest you use python 2.7. I guess the tf.ops are not successfully built, be careful to specify the correct paths to tf versions when you...

hi @fxy2012, either 1.2 or 1.4 should be fine. Yeah, use python 2.7.

Hi @zhangguanghui1 , you are suggested to remove the small ins as well. Use the same function to compute the mean size for each category. https://github.com/Yang7879/3D-BoNet/blob/b4345f28554808f9c07b42bb5487df48c859d89f/main_eval.py#L65

@guanrenxue Data are also available at Baidu Pan: https://pan.baidu.com/s/165IXaA_JISCwGzTUCiuPig 提取码: gbp2

@guanrenxue hmm, the trained models and other intermediate results had been deleted many months ago, due to the server upgrade and the hard drive limitation. Sorry about it, but it's...

@wangyida Sorry, our synthesized dataset does not have a semantic label (e.g., arm, leg, etc..) for each voxel, because ShapeNet does not have semantic labels for different parts of a...

@PranjaLBiswas27 1. There are two steps to convert a raw depth image to a volumetric grid: (1). depth --> point cloud, (2). find the min/max along xyz of the point...

@PranjaLBiswas27 If your depth image is captured by your Kinect v2, then you need to get the correct K matrix for that Kinect. It seems the K matrix you used...