external/cv/mmcv/ops/csrc/pytorch/cpu/voxelization.cpp

// Copyright (c) OpenMMLab. All rights reserved.
#include "pytorch_cpp_helper.hpp"
#include "pytorch_device_registry.hpp"

template <typename T, typename T_int>
void dynamic_voxelize_forward_cpu_kernel(
    const torch::TensorAccessor<T, 2> points,
    torch::TensorAccessor<T_int, 2> coors, const std::vector<float> voxel_size,
    const std::vector<float> coors_range, const std::vector<int> grid_size,
    const int num_points, const int num_features, const int NDim) {
  const int ndim_minus_1 = NDim - 1;
  bool failed = false;
  // int coor[NDim];
  int* coor = new int[NDim]();
  int c;

  for (int i = 0; i < num_points; ++i) {
    failed = false;
    for (int j = 0; j < NDim; ++j) {
      c = floor((points[i][j] - coors_range[j]) / voxel_size[j]);
      // necessary to rm points out of range
      if ((c < 0 || c >= grid_size[j])) {
        failed = true;
        break;
      }
      coor[ndim_minus_1 - j] = c;
    }

    // memcpy and memset will cause problem because of the memory distribution
    // discontinuity of TensorAccessor, so here using loops to replace memcpy
    // or memset
    if (failed) {
      for (int k = 0; k < NDim; ++k) {
        coors[i][k] = -1;
      }
    } else {
      for (int k = 0; k < NDim; ++k) {
        coors[i][k] = coor[k];
      }
    }
  }

  delete[] coor;
  return;
}

template <typename T, typename T_int>
void hard_voxelize_forward_cpu_kernel(
    const torch::TensorAccessor<T, 2> points,
    torch::TensorAccessor<T, 3> voxels, torch::TensorAccessor<T_int, 2> coors,
    torch::TensorAccessor<T_int, 1> num_points_per_voxel,
    torch::TensorAccessor<T_int, 3> coor_to_voxelidx, int& voxel_num,
    const std::vector<float> voxel_size, const std::vector<float> coors_range,
    const std::vector<int> grid_size, const int max_points,
    const int max_voxels, const int num_points, const int num_features,
    const int NDim) {
  // declare a temp coors
  at::Tensor temp_coors = at::zeros(
      {num_points, NDim}, at::TensorOptions().dtype(at::kInt).device(at::kCPU));

  // First use dynamic voxelization to get coors,
  // then check max points/voxels constraints
  dynamic_voxelize_forward_cpu_kernel<T, int>(
      points, temp_coors.accessor<int, 2>(), voxel_size, coors_range, grid_size,
      num_points, num_features, NDim);

  int voxelidx, num;
  auto coor = temp_coors.accessor<int, 2>();

  for (int i = 0; i < num_points; ++i) {
    // T_int* coor = temp_coors.data_ptr<int>() + i * NDim;

    if (coor[i][0] == -1) continue;

    voxelidx = coor_to_voxelidx[coor[i][0]][coor[i][1]][coor[i][2]];

    // record voxel
    if (voxelidx == -1) {
      voxelidx = voxel_num;
      if (max_voxels != -1 && voxel_num >= max_voxels) continue;
      voxel_num += 1;

      coor_to_voxelidx[coor[i][0]][coor[i][1]][coor[i][2]] = voxelidx;
      // memcpy will cause problem because of the memory distribution
      // discontinuity of TensorAccessor, so here using loops to replace memcpy
      for (int k = 0; k < NDim; ++k) {
        coors[voxelidx][k] = coor[i][k];
      }
    }

    // put points into voxel
    num = num_points_per_voxel[voxelidx];
    if (max_points == -1 || num < max_points) {
      // memcpy will cause problem because of the memory distribution
      // discontinuity of TensorAccessor, so here using loops to replace memcpy
      for (int k = 0; k < num_features; ++k) {
        voxels[voxelidx][num][k] = points[i][k];
      }
      num_points_per_voxel[voxelidx] += 1;
    }
  }

  return;
}

void dynamic_voxelize_forward_cpu(const at::Tensor& points, at::Tensor& coors,
                                  const std::vector<float> voxel_size,
                                  const std::vector<float> coors_range,
                                  const int NDim = 3) {
  // check device
  AT_ASSERTM(points.device().is_cpu(), "points must be a CPU tensor");

  std::vector<int> grid_size(NDim);
  const int num_points = points.size(0);
  const int num_features = points.size(1);

  for (int i = 0; i < NDim; ++i) {
    grid_size[i] =
        round((coors_range[NDim + i] - coors_range[i]) / voxel_size[i]);
  }

  // coors, num_points_per_voxel, coor_to_voxelidx are int Tensor
  AT_DISPATCH_FLOATING_TYPES_AND_HALF(
      points.scalar_type(), "dynamic_voxelize_forward_cpu_kernel", [&] {
        dynamic_voxelize_forward_cpu_kernel<scalar_t, int>(
            points.accessor<scalar_t, 2>(), coors.accessor<int, 2>(),
            voxel_size, coors_range, grid_size, num_points, num_features, NDim);
      });
}

int hard_voxelize_forward_cpu(const at::Tensor& points, at::Tensor& voxels,
                              at::Tensor& coors,
                              at::Tensor& num_points_per_voxel,
                              const std::vector<float> voxel_size,
                              const std::vector<float> coors_range,
                              const int max_points, const int max_voxels,
                              const int NDim = 3) {
  // current version tooks about 0.02s_0.03s for one frame on cpu
  // check device
  AT_ASSERTM(points.device().is_cpu(), "points must be a CPU tensor");

  std::vector<int> grid_size(NDim);
  const int num_points = points.size(0);
  const int num_features = points.size(1);

  for (int i = 0; i < NDim; ++i) {
    grid_size[i] =
        round((coors_range[NDim + i] - coors_range[i]) / voxel_size[i]);
  }

  // coors, num_points_per_voxel, coor_to_voxelidx are int Tensor
  // printf("cpu coor_to_voxelidx size: [%d, %d, %d]\n", grid_size[2],
  // grid_size[1], grid_size[0]);
  at::Tensor coor_to_voxelidx =
      -at::ones({grid_size[2], grid_size[1], grid_size[0]}, coors.options());

  int voxel_num = 0;
  AT_DISPATCH_FLOATING_TYPES_AND_HALF(
      points.scalar_type(), "hard_voxelize_forward_cpu_kernel", [&] {
        hard_voxelize_forward_cpu_kernel<scalar_t, int>(
            points.accessor<scalar_t, 2>(), voxels.accessor<scalar_t, 3>(),
            coors.accessor<int, 2>(), num_points_per_voxel.accessor<int, 1>(),
            coor_to_voxelidx.accessor<int, 3>(), voxel_num, voxel_size,
            coors_range, grid_size, max_points, max_voxels, num_points,
            num_features, NDim);
      });

  return voxel_num;
}

int hard_voxelize_forward_impl(const at::Tensor& points, at::Tensor& voxels,
                               at::Tensor& coors,
                               at::Tensor& num_points_per_voxel,
                               const std::vector<float> voxel_size,
                               const std::vector<float> coors_range,
                               const int max_points, const int max_voxels,
                               const int NDim);

void dynamic_voxelize_forward_impl(const at::Tensor& points, at::Tensor& coors,
                                   const std::vector<float> voxel_size,
                                   const std::vector<float> coors_range,
                                   const int NDim);
REGISTER_DEVICE_IMPL(hard_voxelize_forward_impl, CPU,
                     hard_voxelize_forward_cpu);
REGISTER_DEVICE_IMPL(dynamic_voxelize_forward_impl, CPU,
                     dynamic_voxelize_forward_cpu);