resort formal parameters order of hard mining

mmdet/core/bbox/samplers/ohem_sampler.py

@@ -21,23 +21,23 @@ class OHEMSampler(BaseSampler):
         self.bbox_roi_extractor = bbox_roi_extractor
         self.bbox_head = bbox_head
 
-    def hard_mining(self, gallery, assign_result, num_expected, bboxes, feats):
+    def hard_mining(self, inds, num_expected, bboxes, labels, feats):
         # hard mining from the gallery.
         with torch.no_grad():
-            rois = bbox2roi([bboxes[gallery]])
+            rois = bbox2roi([bboxes])
             bbox_feats = self.bbox_roi_extractor(
                 feats[:self.bbox_roi_extractor.num_inputs], rois)
             cls_score, _ = self.bbox_head(bbox_feats)
             loss = self.bbox_head.loss(
                 cls_score=cls_score,
                 bbox_pred=None,
-                labels=assign_result.labels[gallery],
+                labels=labels,
                 label_weights=cls_score.new_ones(cls_score.size(0)),
                 bbox_targets=None,
                 bbox_weights=None,
                 reduce=False)['loss_cls']
             _, topk_loss_inds = loss.topk(num_expected)
-        return gallery[topk_loss_inds]
+        return inds[topk_loss_inds]
 
     def _sample_pos(self,
                     assign_result,
@@ -52,8 +52,8 @@ class OHEMSampler(BaseSampler):
         if pos_inds.numel() <= num_expected:
             return pos_inds
         else:
-            return self.hard_mining(pos_inds, assign_result, num_expected,
-                                    bboxes, feats)
+            return self.hard_mining(pos_inds, num_expected, bboxes[pos_inds],
+                                    assign_result.labels[pos_inds], feats)
 
     def _sample_neg(self,
                     assign_result,
@@ -68,5 +68,5 @@ class OHEMSampler(BaseSampler):
         if len(neg_inds) <= num_expected:
             return neg_inds
         else:
-            return self.hard_mining(neg_inds, assign_result, num_expected,
-                                    bboxes, feats)
+            return self.hard_mining(neg_inds, num_expected, bboxes[neg_inds],
+                                    assign_result.labels[neg_inds], feats)