generate_proposals_op.cu 16.6 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 19
#include <stdio.h>
#include <string>
#include <vector>
#include "cub/cub.cuh"
20
#include "paddle/fluid/framework/mixed_vector.h"
21 22 23 24
#include "paddle/fluid/framework/op_registry.h"
#include "paddle/fluid/memory/memory.h"
#include "paddle/fluid/operators/gather.cu.h"
#include "paddle/fluid/operators/math/math_function.h"
25
#include "paddle/fluid/platform/for_range.h"
26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

namespace paddle {
namespace operators {

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

namespace {

#define DIVUP(m, n) ((m) / (n) + ((m) % (n) > 0))
#define CUDA_1D_KERNEL_LOOP(i, n)                              \
  for (int i = blockIdx.x * blockDim.x + threadIdx.x; i < (n); \
       i += blockDim.x * gridDim.x)

int const kThreadsPerBlock = sizeof(uint64_t) * 8;

42 43 44 45 46 47 48 49
static const double kBBoxClipDefault = std::log(1000.0 / 16.0);

struct RangeInitFunctor {
  int start_;
  int delta_;
  int *out_;
  __device__ void operator()(size_t i) { out_[i] = start_ + i * delta_; }
};
50 51

template <typename T>
52 53 54 55
static void SortDescending(const platform::CUDADeviceContext &ctx,
                           const Tensor &value, Tensor *value_out,
                           Tensor *index_out) {
  int num = static_cast<int>(value.numel());
56 57
  Tensor index_in_t;
  int *idx_in = index_in_t.mutable_data<int>({num}, ctx.GetPlace());
58 59 60
  platform::ForRange<platform::CUDADeviceContext> for_range(ctx, num);
  for_range(RangeInitFunctor{0, 1, idx_in});

61 62 63 64 65 66 67 68
  int *idx_out = index_out->mutable_data<int>({num}, ctx.GetPlace());

  const T *keys_in = value.data<T>();
  T *keys_out = value_out->mutable_data<T>({num}, ctx.GetPlace());

  // Determine temporary device storage requirements
  size_t temp_storage_bytes = 0;
  cub::DeviceRadixSort::SortPairsDescending<T, int>(
69
      nullptr, temp_storage_bytes, keys_in, keys_out, idx_in, idx_out, num);
70 71
  // Allocate temporary storage
  auto place = boost::get<platform::CUDAPlace>(ctx.GetPlace());
72
  auto d_temp_storage = memory::Alloc(place, temp_storage_bytes);
73 74 75

  // Run sorting operation
  cub::DeviceRadixSort::SortPairsDescending<T, int>(
76 77
      d_temp_storage->ptr(), temp_storage_bytes, keys_in, keys_out, idx_in,
      idx_out, num);
78 79 80
}

template <typename T>
81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
struct BoxDecodeAndClipFunctor {
  const T *anchor;
  const T *deltas;
  const T *var;
  const int *index;
  const T *im_info;

  T *proposals;

  BoxDecodeAndClipFunctor(const T *anchor, const T *deltas, const T *var,
                          const int *index, const T *im_info, T *proposals)
      : anchor(anchor),
        deltas(deltas),
        var(var),
        index(index),
        im_info(im_info),
        proposals(proposals) {}

  T bbox_clip_default{static_cast<T>(kBBoxClipDefault)};

  __device__ void operator()(size_t i) {
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117
    int k = index[i] * 4;
    T axmin = anchor[k];
    T aymin = anchor[k + 1];
    T axmax = anchor[k + 2];
    T aymax = anchor[k + 3];

    T w = axmax - axmin + 1.0;
    T h = aymax - aymin + 1.0;
    T cx = axmin + 0.5 * w;
    T cy = aymin + 0.5 * h;

    T dxmin = deltas[k];
    T dymin = deltas[k + 1];
    T dxmax = deltas[k + 2];
    T dymax = deltas[k + 3];

118
    T d_cx, d_cy, d_w, d_h;
119 120 121
    if (var) {
      d_cx = cx + dxmin * w * var[k];
      d_cy = cy + dymin * h * var[k + 1];
122 123
      d_w = exp(Min(dxmax * var[k + 2], bbox_clip_default)) * w;
      d_h = exp(Min(dymax * var[k + 3], bbox_clip_default)) * h;
124 125 126
    } else {
      d_cx = cx + dxmin * w;
      d_cy = cy + dymin * h;
127 128
      d_w = exp(Min(dxmax, bbox_clip_default)) * w;
      d_h = exp(Min(dymax, bbox_clip_default)) * h;
129 130 131 132 133 134 135
    }

    T oxmin = d_cx - d_w * 0.5;
    T oymin = d_cy - d_h * 0.5;
    T oxmax = d_cx + d_w * 0.5 - 1.;
    T oymax = d_cy + d_h * 0.5 - 1.;

136 137 138 139
    proposals[i * 4] = Max(Min(oxmin, im_info[1] - 1.), 0.);
    proposals[i * 4 + 1] = Max(Min(oymin, im_info[0] - 1.), 0.);
    proposals[i * 4 + 2] = Max(Min(oxmax, im_info[1] - 1.), 0.);
    proposals[i * 4 + 3] = Max(Min(oymax, im_info[0] - 1.), 0.);
140
  }
141 142 143 144 145

  __device__ __forceinline__ T Min(T a, T b) const { return a > b ? b : a; }

  __device__ __forceinline__ T Max(T a, T b) const { return a > b ? a : b; }
};
146 147

template <typename T, int BlockSize>
148 149 150
static __global__ void FilterBBoxes(const T *bboxes, const T *im_info,
                                    const T min_size, const int num,
                                    int *keep_num, int *keep) {
151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194
  T im_h = im_info[0];
  T im_w = im_info[1];
  T im_scale = im_info[2];

  int cnt = 0;
  __shared__ int keep_index[BlockSize];

  CUDA_1D_KERNEL_LOOP(i, num) {
    keep_index[threadIdx.x] = -1;
    __syncthreads();

    int k = i * 4;
    T xmin = bboxes[k];
    T ymin = bboxes[k + 1];
    T xmax = bboxes[k + 2];
    T ymax = bboxes[k + 3];

    T w = xmax - xmin + 1.0;
    T h = ymax - ymin + 1.0;
    T cx = xmin + w / 2.;
    T cy = ymin + h / 2.;

    T w_s = (xmax - xmin) / im_scale + 1.;
    T h_s = (ymax - ymin) / im_scale + 1.;

    if (w_s >= min_size && h_s >= min_size && cx <= im_w && cy <= im_h) {
      keep_index[threadIdx.x] = i;
    }
    __syncthreads();
    if (threadIdx.x == 0) {
      int size = (num - i) < BlockSize ? num - i : BlockSize;
      for (int j = 0; j < size; ++j) {
        if (keep_index[j] > -1) {
          keep[cnt++] = keep_index[j];
        }
      }
    }
    __syncthreads();
  }
  if (threadIdx.x == 0) {
    keep_num[0] = cnt;
  }
}

195
static __device__ inline float IoU(const float *a, const float *b) {
196 197 198 199 200 201 202 203 204
  float left = max(a[0], b[0]), right = min(a[2], b[2]);
  float top = max(a[1], b[1]), bottom = min(a[3], b[3]);
  float width = max(right - left + 1, 0.f), height = max(bottom - top + 1, 0.f);
  float inter_s = width * height;
  float s_a = (a[2] - a[0] + 1) * (a[3] - a[1] + 1);
  float s_b = (b[2] - b[0] + 1) * (b[3] - b[1] + 1);
  return inter_s / (s_a + s_b - inter_s);
}

205 206 207
static __global__ void NMSKernel(const int n_boxes,
                                 const float nms_overlap_thresh,
                                 const float *dev_boxes, uint64_t *dev_mask) {
208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248
  const int row_start = blockIdx.y;
  const int col_start = blockIdx.x;

  const int row_size =
      min(n_boxes - row_start * kThreadsPerBlock, kThreadsPerBlock);
  const int col_size =
      min(n_boxes - col_start * kThreadsPerBlock, kThreadsPerBlock);

  __shared__ float block_boxes[kThreadsPerBlock * 4];
  if (threadIdx.x < col_size) {
    block_boxes[threadIdx.x * 4 + 0] =
        dev_boxes[(kThreadsPerBlock * col_start + threadIdx.x) * 4 + 0];
    block_boxes[threadIdx.x * 4 + 1] =
        dev_boxes[(kThreadsPerBlock * col_start + threadIdx.x) * 4 + 1];
    block_boxes[threadIdx.x * 4 + 2] =
        dev_boxes[(kThreadsPerBlock * col_start + threadIdx.x) * 4 + 2];
    block_boxes[threadIdx.x * 4 + 3] =
        dev_boxes[(kThreadsPerBlock * col_start + threadIdx.x) * 4 + 3];
  }
  __syncthreads();

  if (threadIdx.x < row_size) {
    const int cur_box_idx = kThreadsPerBlock * row_start + threadIdx.x;
    const float *cur_box = dev_boxes + cur_box_idx * 4;
    int i = 0;
    uint64_t t = 0;
    int start = 0;
    if (row_start == col_start) {
      start = threadIdx.x + 1;
    }
    for (i = start; i < col_size; i++) {
      if (IoU(cur_box, block_boxes + i * 4) > nms_overlap_thresh) {
        t |= 1ULL << i;
      }
    }
    const int col_blocks = DIVUP(n_boxes, kThreadsPerBlock);
    dev_mask[cur_box_idx * col_blocks + col_start] = t;
  }
}

template <typename T>
249 250 251
static void NMS(const platform::CUDADeviceContext &ctx, const Tensor &proposals,
                const Tensor &sorted_indices, const T nms_threshold,
                Tensor *keep_out) {
252 253 254 255 256 257 258 259 260 261
  int boxes_num = proposals.dims()[0];
  PADDLE_ENFORCE_EQ(boxes_num, sorted_indices.dims()[0]);

  const int col_blocks = DIVUP(boxes_num, kThreadsPerBlock);
  dim3 blocks(DIVUP(boxes_num, kThreadsPerBlock),
              DIVUP(boxes_num, kThreadsPerBlock));
  dim3 threads(kThreadsPerBlock);

  const T *boxes = proposals.data<T>();
  auto place = boost::get<platform::CUDAPlace>(ctx.GetPlace());
262 263 264 265
  framework::Vector<uint64_t> mask(boxes_num * col_blocks);
  NMSKernel<<<blocks, threads>>>(
      boxes_num, nms_threshold, boxes,
      mask.CUDAMutableData(boost::get<platform::CUDAPlace>(ctx.GetPlace())));
266 267 268 269 270 271 272 273 274 275 276 277 278

  std::vector<uint64_t> remv(col_blocks);
  memset(&remv[0], 0, sizeof(uint64_t) * col_blocks);

  std::vector<int> keep_vec;
  int num_to_keep = 0;
  for (int i = 0; i < boxes_num; i++) {
    int nblock = i / kThreadsPerBlock;
    int inblock = i % kThreadsPerBlock;

    if (!(remv[nblock] & (1ULL << inblock))) {
      ++num_to_keep;
      keep_vec.push_back(i);
279
      uint64_t *p = &mask[0] + i * col_blocks;
280 281 282 283 284 285 286
      for (int j = nblock; j < col_blocks; j++) {
        remv[j] |= p[j];
      }
    }
  }
  int *keep = keep_out->mutable_data<int>({num_to_keep}, ctx.GetPlace());
  memory::Copy(place, keep, platform::CPUPlace(), keep_vec.data(),
287 288
               sizeof(int) * num_to_keep, ctx.stream());
  ctx.Wait();
289 290 291
}

template <typename T>
292
static std::pair<Tensor, Tensor> ProposalForOneImage(
293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310
    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());
311 312 313 314 315 316 317

  {
    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>()});
  }
318 319 320 321 322 323

  // 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);
324
  auto stream = ctx.stream();
325 326 327 328 329 330
  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;
  const auto gpu_place = boost::get<platform::CUDAPlace>(ctx.GetPlace());
  memory::Copy(platform::CPUPlace(), &keep_num, gpu_place,
331 332
               keep_num_t.data<int>(), sizeof(int), ctx.stream());
  ctx.Wait();
333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368
  keep_index.Resize({keep_num});

  Tensor scores_filter, proposals_filter;
  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");
369 370 371 372
    auto anchors = GET_DATA_SAFELY(context.Input<Tensor>("Anchors"), "Input",
                                   "Anchors", "GenerateProposals");
    auto variances = GET_DATA_SAFELY(context.Input<Tensor>("Variances"),
                                     "Input", "Variances", "GenerateProposals");
373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407

    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");
    PADDLE_ENFORCE_GE(eta, 1., "Not support adaptive NMS.");

    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);

408 409
    anchors.Resize({anchors.numel() / 4, 4});
    variances.Resize({variances.numel() / 4, 4});
410 411 412 413 414 415 416 417

    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>();

D
Dang Qingqing 已提交
418
    auto place = boost::get<platform::CUDAPlace>(dev_ctx.GetPlace());
F
FDInSky 已提交
419
    auto cpu_place = platform::CPUPlace();
420 421 422

    int64_t num_proposals = 0;
    std::vector<size_t> offset(1, 0);
F
FDInSky 已提交
423 424
    std::vector<int64_t> tmp_lod;

425 426 427 428 429 430 431 432 433
    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 =
434
          ProposalForOneImage<T>(dev_ctx, im_info_slice, anchors, variances,
435 436 437
                                 bbox_deltas_slice, scores_slice, pre_nms_top_n,
                                 post_nms_top_n, nms_thresh, min_size, eta);

438 439
      Tensor &proposals = box_score_pair.first;
      Tensor &scores = box_score_pair.second;
440 441

      memory::Copy(place, rpn_rois_data + num_proposals * 4, place,
442 443
                   proposals.data<T>(), sizeof(T) * proposals.numel(),
                   dev_ctx.stream());
444
      memory::Copy(place, rpn_roi_probs_data + num_proposals, place,
445 446 447
                   scores.data<T>(), sizeof(T) * scores.numel(),
                   dev_ctx.stream());
      dev_ctx.Wait();
448 449
      num_proposals += proposals.dims()[0];
      offset.emplace_back(num_proposals);
F
FDInSky 已提交
450 451 452 453 454 455 456 457 458
      tmp_lod.push_back(num_proposals);
    }
    if (context.HasOutput("RpnRoisLod")) {
      auto *rpn_rois_lod = context.Output<Tensor>("RpnRoisLod");
      rpn_rois_lod->mutable_data<int64_t>({num}, context.GetPlace());
      int64_t *lod_data = rpn_rois_lod->data<int64_t>();
      memory::Copy(place, lod_data, cpu_place, &tmp_lod[0],
                   sizeof(int64_t) * num, dev_ctx.stream());
      rpn_rois_lod->Resize({num});
459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475
    }
    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>);