experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Internal change

Browse Source 
PiperOrigin-RevId: 358086483
Change-Id: I3226271fc2065713cdf59ae14110db8d738a8a99
This commit is contained in:
A. Unique TensorFlower 2021-02-17 19:32:12 -08:00 committed by TensorFlower Gardener
parent 630a18b20f
commit 72e1ddc763
1 changed files with 2 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
tensorflow/lite/kernels/detection_postprocess.cc
View File

@ -367,16 +367,6 @@ void DecreasingPartialArgSort(const float* values, int num_values,
[&values](const int i, const int j) { return values[i] > values[j]; });
}
void DecreasingArgSort(const float* values, int num_values, int* indices) {
std::iota(indices, indices + num_values, 0);
// We want here a stable sort, in order to get completely defined output.
// In this way TFL and TFLM can be bit-exact.
std::stable_sort(
indices, indices + num_values,
[&values](const int i, const int j) { return values[i] > values[j]; });
}
void SelectDetectionsAboveScoreThreshold(const std::vector<float>& values,
const float threshold,
std::vector<float>* keep_values,
@ -461,8 +451,8 @@ TfLiteStatus NonMaxSuppressionSingleClassHelper(
int num_scores_kept = keep_scores.size();
std::vector<int> sorted_indices;
sorted_indices.resize(num_scores_kept);
DecreasingArgSort(keep_scores.data(), num_scores_kept, sorted_indices.data());
DecreasingPartialArgSort(keep_scores.data(), num_scores_kept, num_scores_kept,
sorted_indices.data());
const int num_boxes_kept = num_scores_kept;
const int output_size = std::min(num_boxes_kept, max_detections);
selected->clear();