Use int64_t to track selected_boxes_inside_class

This cleans up the compute_nms signature as well as stops using additional
memory by not storing every pair result twice that just gets cleared per run each shape_for_each()

src/include/migraphx/op/nonmaxsuppression.hpp

@@ -260,7 +260,6 @@ struct nonmaxsuppression
     std::size_t compute_nms(Output output,
                             Boxes boxes,
                             Scores scores,
-                            const shape& max_output_shape,
                             std::size_t max_output_boxes_per_class,
                             double iou_threshold,
                             double score_threshold) const
@@ -271,9 +270,7 @@ struct nonmaxsuppression
         const auto num_classes = lens[1];
         const auto num_boxes   = lens[2];
         // boxes of a class with NMS applied [score, index]
-        std::vector<std::pair<double, int64_t>> selected_boxes_inside_class;
         std::vector<int64_t> selected_indices;
-        selected_boxes_inside_class.reserve(max_output_shape.elements());
         // iterate over batches and classes
         shape comp_s{shape::double_type, {num_batches, num_classes}};
         shape_for_each(comp_s, [&](auto idx) {
@@ -284,20 +281,21 @@ struct nonmaxsuppression
             // iterator to first value of this batch
             auto batch_boxes_start = boxes.begin() + batch_idx * num_boxes * 4;
             auto boxes_heap = filter_boxes_by_score(scores_start, num_boxes, score_threshold);
-            selected_boxes_inside_class.clear();
+            int64_t selected_boxes_inside_class = 0;
             while(not boxes_heap.empty() &&
-                  selected_boxes_inside_class.size() < max_output_boxes_per_class)
+                  selected_boxes_inside_class < max_output_boxes_per_class)
             {
                 // select next top scorer box and remove any boxes from boxes_heap that exceeds IOU
                 // threshold with the selected box
                 const auto next_top_score = boxes_heap.top();
                 auto next_box             = batch_box(batch_boxes_start, next_top_score.second);
+                auto next_box_idx         = next_top_score.second;
                 boxes_heap.pop();
 
-                selected_boxes_inside_class.push_back(next_top_score);
+                selected_boxes_inside_class++;
                 selected_indices.push_back(batch_idx);
                 selected_indices.push_back(class_idx);
-                selected_indices.push_back(next_top_score.second);
+                selected_indices.push_back(next_box_idx);
                 std::priority_queue<std::pair<double, int64_t>> remainder_boxes;
                 while(not boxes_heap.empty())
                 {