generate_proposals_op.cu 9.0 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14
/* Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. */

Y
Yu Yang 已提交
15
#include <paddle/fluid/memory/allocation/allocator.h>
16 17 18
#include <stdio.h>
#include <string>
#include <vector>
19
#include "paddle/fluid/framework/mixed_vector.h"
20 21
#include "paddle/fluid/framework/op_registry.h"
#include "paddle/fluid/memory/memory.h"
22
#include "paddle/fluid/operators/detection/bbox_util.cu.h"
23 24 25 26 27 28 29 30 31 32
#include "paddle/fluid/operators/math/math_function.h"

namespace paddle {
namespace operators {

using Tensor = framework::Tensor;
using LoDTensor = framework::LoDTensor;

namespace {
template <typename T>
33
static std::pair<Tensor, Tensor> ProposalForOneImage(
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
    const platform::CUDADeviceContext &ctx, const Tensor &im_info,
    const Tensor &anchors, const Tensor &variances,
    const Tensor &bbox_deltas,  // [M, 4]
    const Tensor &scores,       // [N, 1]
    int pre_nms_top_n, int post_nms_top_n, float nms_thresh, float min_size,
    float eta) {
  // 1. pre nms
  Tensor scores_sort, index_sort;
  SortDescending<T>(ctx, scores, &scores_sort, &index_sort);
  int num = scores.numel();
  int pre_nms_num = (pre_nms_top_n <= 0 || pre_nms_top_n > num) ? scores.numel()
                                                                : pre_nms_top_n;
  scores_sort.Resize({pre_nms_num, 1});
  index_sort.Resize({pre_nms_num, 1});

  // 2. box decode and clipping
  Tensor proposals;
  proposals.mutable_data<T>({pre_nms_num, 4}, ctx.GetPlace());
52 53 54 55 56 57 58

  {
    platform::ForRange<platform::CUDADeviceContext> for_range(ctx, pre_nms_num);
    for_range(BoxDecodeAndClipFunctor<T>{
        anchors.data<T>(), bbox_deltas.data<T>(), variances.data<T>(),
        index_sort.data<int>(), im_info.data<T>(), proposals.data<T>()});
  }
59 60 61 62 63 64

  // 3. filter
  Tensor keep_index, keep_num_t;
  keep_index.mutable_data<int>({pre_nms_num}, ctx.GetPlace());
  keep_num_t.mutable_data<int>({1}, ctx.GetPlace());
  min_size = std::max(min_size, 1.0f);
65
  auto stream = ctx.stream();
66 67 68 69
  FilterBBoxes<T, 512><<<1, 512, 0, stream>>>(
      proposals.data<T>(), im_info.data<T>(), min_size, pre_nms_num,
      keep_num_t.data<int>(), keep_index.data<int>());
  int keep_num;
70
  const auto gpu_place = BOOST_GET_CONST(platform::CUDAPlace, ctx.GetPlace());
71
  memory::Copy(platform::CPUPlace(), &keep_num, gpu_place,
72 73
               keep_num_t.data<int>(), sizeof(int), ctx.stream());
  ctx.Wait();
74 75 76
  keep_index.Resize({keep_num});

  Tensor scores_filter, proposals_filter;
77 78 79 80 81 82 83 84 85
  // Handle the case when there is no keep index left
  if (keep_num == 0) {
    math::SetConstant<platform::CUDADeviceContext, T> set_zero;
    proposals_filter.mutable_data<T>({1, 4}, ctx.GetPlace());
    scores_filter.mutable_data<T>({1, 1}, ctx.GetPlace());
    set_zero(ctx, &proposals_filter, static_cast<T>(0));
    set_zero(ctx, &scores_filter, static_cast<T>(0));
    return std::make_pair(proposals_filter, scores_filter);
  }
86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118
  proposals_filter.mutable_data<T>({keep_num, 4}, ctx.GetPlace());
  scores_filter.mutable_data<T>({keep_num, 1}, ctx.GetPlace());
  GPUGather<T>(ctx, proposals, keep_index, &proposals_filter);
  GPUGather<T>(ctx, scores_sort, keep_index, &scores_filter);

  if (nms_thresh <= 0) {
    return std::make_pair(proposals_filter, scores_filter);
  }

  // 4. nms
  Tensor keep_nms;
  NMS<T>(ctx, proposals_filter, keep_index, nms_thresh, &keep_nms);
  if (post_nms_top_n > 0 && post_nms_top_n < keep_nms.numel()) {
    keep_nms.Resize({post_nms_top_n});
  }

  Tensor scores_nms, proposals_nms;
  proposals_nms.mutable_data<T>({keep_nms.numel(), 4}, ctx.GetPlace());
  scores_nms.mutable_data<T>({keep_nms.numel(), 1}, ctx.GetPlace());
  GPUGather<T>(ctx, proposals_filter, keep_nms, &proposals_nms);
  GPUGather<T>(ctx, scores_filter, keep_nms, &scores_nms);

  return std::make_pair(proposals_nms, scores_nms);
}
}  // namespace

template <typename DeviceContext, typename T>
class CUDAGenerateProposalsKernel : public framework::OpKernel<T> {
 public:
  void Compute(const framework::ExecutionContext &context) const override {
    auto *scores = context.Input<Tensor>("Scores");
    auto *bbox_deltas = context.Input<Tensor>("BboxDeltas");
    auto *im_info = context.Input<Tensor>("ImInfo");
119 120 121 122
    auto anchors = GET_DATA_SAFELY(context.Input<Tensor>("Anchors"), "Input",
                                   "Anchors", "GenerateProposals");
    auto variances = GET_DATA_SAFELY(context.Input<Tensor>("Variances"),
                                     "Input", "Variances", "GenerateProposals");
123 124 125 126 127 128 129 130 131

    auto *rpn_rois = context.Output<LoDTensor>("RpnRois");
    auto *rpn_roi_probs = context.Output<LoDTensor>("RpnRoiProbs");

    int pre_nms_top_n = context.Attr<int>("pre_nms_topN");
    int post_nms_top_n = context.Attr<int>("post_nms_topN");
    float nms_thresh = context.Attr<float>("nms_thresh");
    float min_size = context.Attr<float>("min_size");
    float eta = context.Attr<float>("eta");
132 133 134 135 136
    PADDLE_ENFORCE_GE(eta, 1.,
                      platform::errors::InvalidArgument(
                          "Not support adaptive NMS. The attribute 'eta' "
                          "should not less than 1. But received eta=[%d]",
                          eta));
137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161

    auto &dev_ctx = context.template device_context<DeviceContext>();

    auto scores_dim = scores->dims();
    int64_t num = scores_dim[0];
    int64_t c_score = scores_dim[1];
    int64_t h_score = scores_dim[2];
    int64_t w_score = scores_dim[3];

    auto bbox_dim = bbox_deltas->dims();
    int64_t c_bbox = bbox_dim[1];
    int64_t h_bbox = bbox_dim[2];
    int64_t w_bbox = bbox_dim[3];

    Tensor bbox_deltas_swap, scores_swap;
    bbox_deltas_swap.mutable_data<T>({num, h_bbox, w_bbox, c_bbox},
                                     dev_ctx.GetPlace());
    scores_swap.mutable_data<T>({num, h_score, w_score, c_score},
                                dev_ctx.GetPlace());

    math::Transpose<DeviceContext, T, 4> trans;
    std::vector<int> axis = {0, 2, 3, 1};
    trans(dev_ctx, *bbox_deltas, &bbox_deltas_swap, axis);
    trans(dev_ctx, *scores, &scores_swap, axis);

162 163
    anchors.Resize({anchors.numel() / 4, 4});
    variances.Resize({variances.numel() / 4, 4});
164 165 166 167 168 169 170 171

    rpn_rois->mutable_data<T>({bbox_deltas->numel() / 4, 4},
                              context.GetPlace());
    rpn_roi_probs->mutable_data<T>({scores->numel(), 1}, context.GetPlace());

    T *rpn_rois_data = rpn_rois->data<T>();
    T *rpn_roi_probs_data = rpn_roi_probs->data<T>();

172
    auto place = BOOST_GET_CONST(platform::CUDAPlace, dev_ctx.GetPlace());
F
FDInSky 已提交
173
    auto cpu_place = platform::CPUPlace();
174 175 176

    int64_t num_proposals = 0;
    std::vector<size_t> offset(1, 0);
177
    std::vector<int> tmp_num;
F
FDInSky 已提交
178

179 180 181 182 183 184 185 186 187
    for (int64_t i = 0; i < num; ++i) {
      Tensor im_info_slice = im_info->Slice(i, i + 1);
      Tensor bbox_deltas_slice = bbox_deltas_swap.Slice(i, i + 1);
      Tensor scores_slice = scores_swap.Slice(i, i + 1);

      bbox_deltas_slice.Resize({h_bbox * w_bbox * c_bbox / 4, 4});
      scores_slice.Resize({h_score * w_score * c_score, 1});

      std::pair<Tensor, Tensor> box_score_pair =
188
          ProposalForOneImage<T>(dev_ctx, im_info_slice, anchors, variances,
189 190 191
                                 bbox_deltas_slice, scores_slice, pre_nms_top_n,
                                 post_nms_top_n, nms_thresh, min_size, eta);

192 193
      Tensor &proposals = box_score_pair.first;
      Tensor &scores = box_score_pair.second;
194 195

      memory::Copy(place, rpn_rois_data + num_proposals * 4, place,
196 197
                   proposals.data<T>(), sizeof(T) * proposals.numel(),
                   dev_ctx.stream());
198
      memory::Copy(place, rpn_roi_probs_data + num_proposals, place,
199 200 201
                   scores.data<T>(), sizeof(T) * scores.numel(),
                   dev_ctx.stream());
      dev_ctx.Wait();
202 203
      num_proposals += proposals.dims()[0];
      offset.emplace_back(num_proposals);
204
      tmp_num.push_back(proposals.dims()[0]);
F
FDInSky 已提交
205
    }
206 207 208 209 210 211 212
    if (context.HasOutput("RpnRoisNum")) {
      auto *rpn_rois_num = context.Output<Tensor>("RpnRoisNum");
      rpn_rois_num->mutable_data<int>({num}, context.GetPlace());
      int *num_data = rpn_rois_num->data<int>();
      memory::Copy(place, num_data, cpu_place, &tmp_num[0], sizeof(int) * num,
                   dev_ctx.stream());
      rpn_rois_num->Resize({num});
213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229
    }
    framework::LoD lod;
    lod.emplace_back(offset);
    rpn_rois->set_lod(lod);
    rpn_roi_probs->set_lod(lod);
    rpn_rois->Resize({num_proposals, 4});
    rpn_roi_probs->Resize({num_proposals, 1});
  }
};

}  // namespace operators
}  // namespace paddle

namespace ops = paddle::operators;
REGISTER_OP_CUDA_KERNEL(generate_proposals,
                        ops::CUDAGenerateProposalsKernel<
                            paddle::platform::CUDADeviceContext, float>);