Skip to content

cnms opt #42

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Feb 9, 2021
Merged

cnms opt #42

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 12 additions & 2 deletions tensorflow/core/kernels/image/non_max_suppression_op.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,7 +24,6 @@ limitations under the License.
#include <queue>
#include <vector>

#include "third_party/eigen3/unsupported/Eigen/CXX11/Tensor"
#include "tensorflow/core/framework/bounds_check.h"
#include "tensorflow/core/framework/op_kernel.h"
#include "tensorflow/core/framework/register_types.h"
Expand All @@ -33,6 +32,7 @@ limitations under the License.
#include "tensorflow/core/framework/types.h"
#include "tensorflow/core/lib/core/status.h"
#include "tensorflow/core/platform/logging.h"
#include "third_party/eigen3/unsupported/Eigen/CXX11/Tensor"

namespace tensorflow {
namespace {
Expand Down Expand Up @@ -963,14 +963,24 @@ class CombinedNonMaxSuppressionOp : public OpKernel {
errors::InvalidArgument("iou_threshold must be in [0, 1]"));
int num_boxes = 0;
const int num_classes = scores.dim_size(2);
// out_per_class is a heuristic below to limit the per_class max detections
// so that the amount of unnecessary IOU computations are reduced. Algorthm
// picks the top k detections out of a possible k*num_classes detections.
// The heuristic below assumes uniform dist of objects among classes
// and adds a buffer of a constant max_output_size/2 to accomodate for
// non-uniformity.
int guard_band = max_output_size.scalar<int>()()/4;
int num_classes_p = num_classes > 0 ? num_classes : 1;
int out_per_class = !pad_per_class_ ? std::ceil(static_cast<float>(max_total_size_per_batch)/num_classes_p) + guard_band : max_output_size.scalar<int>()();
out_per_class = std::min(out_per_class, max_output_size.scalar<int>()());
ParseAndCheckCombinedNMSBoxSizes(context, boxes, &num_boxes, num_classes);
CheckCombinedNMSScoreSizes(context, num_boxes, scores);

if (!context->status().ok()) {
return;
}
BatchedNonMaxSuppressionOp(context, boxes, scores, num_boxes,
max_size_per_class, max_total_size_per_batch,
out_per_class, max_total_size_per_batch,
score_threshold_val, iou_threshold_val,
pad_per_class_, clip_boxes_);
}
Expand Down