/home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-roccv/checkouts/latest/include/kernels/host/non_max_suppression_host.hpp Source File

/home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-roccv/checkouts/latest/include/kernels/host/non_max_suppression_host.hpp Source File#

3 min read time

Applies to Linux

rocCV: /home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-roccv/checkouts/latest/include/kernels/host/non_max_suppression_host.hpp Source File

Go to the documentation of this file.

 /*
  * Copyright (c) 2025 Advanced Micro Devices, Inc. All rights reserved.
  * Permission is hereby granted, free of charge, to any person obtaining a copy
  * of this software and associated documentation files (the "Software"), to deal
  * in the Software without restriction, including without limitation the rights
  * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
  * copies of the Software, and to permit persons to whom the Software is
  * furnished to do so, subject to the following conditions:
  *
  * The above copyright notice and this permission notice shall be included in
  * all copies or substantial portions of the Software.
  *
  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
  * THE SOFTWARE.
  */
  
 #pragma once
  
 #include <hip/hip_runtime.h>
  
 #include "core/wrappers/generic_tensor_wrapper.hpp"
 #include "kernels/common/non_max_suppression_helpers.hpp"
  
 namespace Kernels {
 namespace Host {
 inline void non_maximum_suppression(roccv::GenericTensorWrapper<short4> input,
                                     roccv::GenericTensorWrapper<uint8_t> output,
                                     roccv::GenericTensorWrapper<float> scores, int numBoxes, float scoresThreshold,
                                     float iouThreshold) {
     for (int64_t batchIdx = 0; batchIdx < input.shape[0]; batchIdx++) {
         for (int64_t boxAIdx = 0; boxAIdx < numBoxes; boxAIdx++) {
             const float scoreA = scores.at(batchIdx, boxAIdx);
             uint8_t& dst = output.at(batchIdx, boxAIdx);
  
             if (scoreA < scoresThreshold) {
                 dst = 0u;
                 continue;
             }
  
             const short4 boxA = input.at(batchIdx, boxAIdx);
             bool discard = false;
  
             for (int boxBIdx = 0; boxBIdx < numBoxes; boxBIdx++) {
                 if (boxBIdx == boxAIdx) continue;
  
                 const short4 boxB = input.at(batchIdx, boxBIdx);
                 if (ComputeIoU(boxA, boxB) > iouThreshold) {
                     const float scoreB = scores.at(batchIdx, boxBIdx);
                     if (scoreA < scoreB || (scoreA == scoreB && ComputeArea(boxA) < ComputeArea(boxB))) {
                         discard = true;
                         break;
                     }
                 }
             }
  
             dst = discard ? 0u : 1u;
         }
     }
 }
 }  // namespace Host
 }  // namespace Kernels