Test_csv infile

[liuxj1]

Hello! When restoring the complete point cloud in Test_csv, why does the restoration effect of the customized residual point cloud (I chose a fixed point direction, and the 512 points closest to the selected direction point in the 2048 point cloud are removed), be very bad? Are there any requirements for using Test_csv input?

[liuxj1]

OK, this issue has been resolved for now. But I still have a question: does the completed point cloud only target residual point clouds that are lost in the five directions set when training the model? Will the effect be poor for residual point clouds that are randomly lost?

[xiaoxutongxu]

@liuxj1 请问如何将自定义残差点云按5个方向去掉点数？

[liuxj1]

源码中定义残缺点云方式： choice = [torch.Tensor([1,0,0]),torch.Tensor([0,0,1]),torch.Tensor([1,0,1]),torch.Tensor([-1,0,0]),torch.Tensor([-1,1,0])] for m in range(batch_size): index = random.sample(choice,1)#Random choose one of the viewpoint distance_list = [] p_center = index[0] for n in range(opt.pnum): distance_list.append(distance_squre(real_point[m,0,n],p_center)) distance_order = sorted(enumerate(distance_list), key = lambda x:x[1])

                for sp in range(opt.crop_point_num):
                    input_cropped1.data[m,0,distance_order[sp][0]] = torch.FloatTensor([0,0,0])
                    real_center.data[m,0,sp] = real_point[m,0,distance_order[sp][0]]，在自己测试模型时，自定义残缺点云也应该要满足这种定义方式。前面choice就是训练中定义的五个缺失视点。随机选择一个视点，distance_squre(real_point[m,0,n],p_center)计算点云距离该视点的距离，最近的512点作为丢失点。记住需要提前归一化点云。