init LOCE

fcjian · Aug 18, 2021 · c5c523a · c5c523a
1 parent 2930b4e
commit c5c523a
Showing 1 changed file with 10 additions and 13 deletions.
diff --git a/mmdet/models/roi_heads/loce_roi_head.py b/mmdet/models/roi_heads/loce_roi_head.py
@@ -199,11 +199,11 @@ def _get_feat_for_memory(self, x, gt_bboxes, gt_labels, img_metas):
             x[:self.bbox_roi_extractor.num_inputs], rois)
         bbox_feats = bbox_feats.detach()  # 0
 
-        gt_labels, _, reg_targets, _ = self.bbox_head.get_targets_for_memory(proposal_list, neg_proposal_list,
+        bbox_labels, _, bbox_targets, _ = self.bbox_head.get_targets_for_memory(proposal_list, neg_proposal_list,
                                                              gt_bbox_list, gt_label_list,
                                                              self.train_cfg)
 
-        return bbox_feats, gt_labels, reg_targets
+        return bbox_feats, bbox_labels, bbox_targets
 
     def _compute_batch_mean_score(self, cls_score, sampling_results, gt_labels, selected_labels):
         # batch mean score for current sample (non queue sampling samples)
@@ -212,15 +212,12 @@ def _compute_batch_mean_score(self, cls_score, sampling_results, gt_labels, sele
         batch_gt_labels = []
         batch_mean_scores = []
         for img_ind, sampling_results_img in enumerate(sampling_results):
-            try:
-                for gt_ind, gt_label in enumerate(gt_labels[img_ind]):
-                    if (sampling_results_img.pos_assigned_gt_inds == gt_ind).sum() > 0:
-                        score = scores[self.bbox_sampler.num * img_ind:self.bbox_sampler.num * img_ind + len(sampling_results_img.pos_assigned_gt_inds),
-                            gt_label][sampling_results_img.pos_assigned_gt_inds == gt_ind]
-                        batch_gt_labels.append(gt_label.unsqueeze(0))
-                        batch_mean_scores.append(score.mean().unsqueeze(0))
-            except:
-                print(gt_labels)
+            for gt_ind, gt_label in enumerate(gt_labels[img_ind]):
+                if (sampling_results_img.pos_assigned_gt_inds == gt_ind).sum() > 0:
+                    score = scores[self.bbox_sampler.num * img_ind:self.bbox_sampler.num * img_ind + len(sampling_results_img.pos_assigned_gt_inds),
+                        gt_label][sampling_results_img.pos_assigned_gt_inds == gt_ind]
+                    batch_gt_labels.append(gt_label.unsqueeze(0))
+                    batch_mean_scores.append(score.mean().unsqueeze(0))
 
         # batch mean score for selected queue samples
         selected_length = len(selected_labels)
@@ -263,8 +260,8 @@ def _bbox_forward_train(self, x, sampling_results, gt_bboxes, gt_labels,
             rank = -1
 
         # for memory-augmented feature sampling
-        bbox_feats_for_memory, gt_labels_for_memory, reg_targets_for_memory = self._get_feat_for_memory(x, gt_bboxes, gt_labels, img_metas)
-        self.mfs.enqueue_dequeue(bbox_feats_for_memory, gt_labels_for_memory, reg_targets_for_memory)
+        bbox_feats, bbox_labels, bbox_targets = self._get_feat_for_memory(x, gt_bboxes, gt_labels, img_metas)
+        self.mfs.enqueue_dequeue(bbox_feats, bbox_labels, bbox_targets)
         selectd_bbox_feat, selectd_labels, selectd_reg_targets, selectd_cls_weight, selectd_reg_weight = \
                         self.mfs.probabilistic_sampler(self.mean_score)