[dev] s2anet models (#3803)

* s2anet update

* remove np anchor

configs/dota/README.md

@@ -129,9 +129,10 @@ python3.7 tools/infer.py -c configs/dota/s2anet_1x_dota.yml -o weights=./weights
 
 ### S2ANet模型
 
-|     模型     | GPU个数  |  Conv类型  |   mAP    |   模型下载   |   配置文件   |
-|:-----------:|:-------:|:----------:|:--------:| :----------:| :---------: |
-|   S2ANet    |    8    |   Conv     |   71.42  |  [model](https://paddledet.bj.bcebos.com/models/s2anet_conv_1x_dota.pdparams) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/dota/s2anet_conv_1x_dota.yml)                   |
+|     模型     |  Conv类型  |   mAP    |   模型下载   |   配置文件   |
+|:-----------:|:----------:|:--------:| :----------:| :---------: |
+|   S2ANet    |   Conv     |   71.42  |  [model](https://paddledet.bj.bcebos.com/models/s2anet_conv_1x_dota.pdparams) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/dota/s2anet_conv_1x_dota.yml)                   |
+|   S2ANet    |  AlignConv |   74.0   |  [model](https://paddledet.bj.bcebos.com/models/s2anet_alignconv_2x_dota.pdparams) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/dota/s2anet_alignconv_2x_dota.yml)                   |
 
 **注意：**这里使用`multiclass_nms`，与原作者使用nms略有不同，精度相比原始论文中高0.15 (71.27-->71.42)。
 

configs/dota/_base_/s2anet.yml

@@ -53,4 +53,3 @@ S2ANetBBoxPostProcess:
     score_threshold: 0.05
     nms_threshold: 0.1
     normalized: False
-    #background_label: -1

configs/dota/_base_/s2anet_optimizer_2x.yml

+epoch: 24
+
+LearningRate:
+  base_lr: 0.005
+  schedulers:
+  - !PiecewiseDecay
+    gamma: 0.1
+    milestones: [14, 20]
+  - !LinearWarmup
+    start_factor: 0.3333333333333333
+    steps: 1000
+
+OptimizerBuilder:
+  optimizer:
+    momentum: 0.9
+    type: Momentum
+  regularizer:
+    factor: 0.0001
+    type: L2
+  clip_grad_by_norm: 35

configs/dota/s2anet_1x_spine.yml

@@ -24,5 +24,7 @@ S2ANetHead:
   align_conv_type: 'AlignConv'  # AlignConv Conv
   align_conv_size: 3
   use_sigmoid_cls: True
-  reg_loss_weight: [1.0, 1.0, 1.0, 1.0, 1.05]                                                                  
-  cls_loss_weight: [1.05, 1.0] 
+  reg_loss_weight: [1.0, 1.0, 1.0, 1.0, 1.05]
+  cls_loss_weight: [1.05, 1.0]
+  reg_loss_type: gwd
+  use_paddle_anchor: False

configs/dota/s2anet_1x_dota.yml → configs/dota/s2anet_alignconv_2x_dota.yml

-it _BASE_: [
+_BASE_: [
   '../datasets/dota.yml',
   '../runtime.yml',
-  '_base_/s2anet_optimizer_1x.yml',
+  '_base_/s2anet_optimizer_2x.yml',
   '_base_/s2anet.yml',
   '_base_/s2anet_reader.yml',
 ]
-weights: output/s2anet_1x_dota/model_final
+pretrain_weights: https://paddledet.bj.bcebos.com/models/pretrained/ResNet50_vd_ssld_v2_pretrained.pdparams
+
+weights: output/s2anet_alignconv_2x_dota/model_final
 
 S2ANetHead:
   anchor_strides: [8, 16, 32, 64, 128]
@@ -19,5 +21,6 @@ S2ANetHead:
   align_conv_type: 'AlignConv'  # AlignConv Conv
   align_conv_size: 3
   use_sigmoid_cls: True
-  reg_loss_weight: [1.0, 1.0, 1.0, 1.0, 1.1]
-  cls_loss_weight: [1.1, 1.05]
+  reg_loss_weight: [1.0, 1.0, 1.0, 1.0, 1.05]
+  cls_loss_weight: [1.05, 1.0]
+  #reg_loss_type: 'l1' # 'l1' 'gwd'

configs/dota/s2anet_conv_1x_dota.yml

@@ -7,6 +7,13 @@ _BASE_: [
 ]
 weights: output/s2anet_1x_dota/model_final
 
+ResNet:
+  depth: 50
+  variant: b
+  norm_type: bn
+  return_idx: [1,2,3]
+  num_stages: 4
+
 S2ANetHead:
   anchor_strides: [8, 16, 32, 64, 128]
   anchor_scales: [4]

ppdet/modeling/heads/s2anet_head.py

@@ -102,8 +102,7 @@ class S2ANetAnchorGenerator(nn.Layer):
         valid_xx, valid_yy = self._meshgrid(valid_x, valid_y)
         valid = valid_xx & valid_yy
         valid = paddle.reshape(valid, [-1, 1])
-        valid = paddle.expand(valid,
-            [-1, self.num_base_anchors]).reshape([-1])
+        valid = paddle.expand(valid, [-1, self.num_base_anchors]).reshape([-1])
         return valid
 
 
@@ -179,9 +178,12 @@ class AlignConv(nn.Layer):
         offset_x = x_anchor - x_conv
         offset_y = y_anchor - y_conv
         offset = paddle.stack([offset_y, offset_x], axis=-1)
-        offset = paddle.reshape(offset, [feat_h * feat_w, self.kernel_size * self.kernel_size * 2])                    
+        offset = paddle.reshape(
+            offset, [feat_h * feat_w, self.kernel_size * self.kernel_size * 2])
         offset = paddle.transpose(offset, [1, 0])
-        offset = paddle.reshape(offset, [1, self.kernel_size * self.kernel_size * 2, feat_h, feat_w]) 
+        offset = paddle.reshape(
+            offset,
+            [1, self.kernel_size * self.kernel_size * 2, feat_h, feat_w])
         return offset
 
     def forward(self, x, refine_anchors, featmap_size, stride):
@@ -260,8 +262,8 @@ class S2ANetHead(nn.Layer):
         # anchor
         self.anchor_generators = []
         for anchor_base in self.anchor_base_sizes:
-                self.anchor_generators.append(
-                    S2ANetAnchorGenerator(anchor_base, anchor_scales,
+            self.anchor_generators.append(
+                S2ANetAnchorGenerator(anchor_base, anchor_scales,
                                           anchor_ratios))
 
         self.anchor_generators = nn.LayerList(self.anchor_generators)
@@ -440,8 +442,7 @@ class S2ANetHead(nn.Layer):
 
             init_anchors = paddle.to_tensor(init_anchors, dtype='float32')
             NA = featmap_size[0] * featmap_size[1]
-            init_anchors = paddle.reshape(
-                init_anchors, [NA, 4])
+            init_anchors = paddle.reshape(init_anchors, [NA, 4])
             init_anchors = self.rect2rbox(init_anchors)
             self.base_anchors_list.append(init_anchors)
 
@@ -474,18 +475,19 @@ class S2ANetHead(nn.Layer):
             # [N, CLS, H, W] --> [N, H, W, CLS]
             odm_cls_score = odm_cls_score.transpose([0, 2, 3, 1])
             odm_cls_score_shape = odm_cls_score.shape
-            odm_cls_score_reshape = paddle.reshape(
-                odm_cls_score,
-                [odm_cls_score_shape[0], odm_cls_score_shape[1] * odm_cls_score_shape[2], self.cls_out_channels])
+            odm_cls_score_reshape = paddle.reshape(odm_cls_score, [
+                odm_cls_score_shape[0], odm_cls_score_shape[1] *
+                odm_cls_score_shape[2], self.cls_out_channels
+            ])
 
             odm_cls_branch_list.append(odm_cls_score_reshape)
 
             odm_bbox_pred = self.odm_reg(odm_reg_feat)
             # [N, 5, H, W] --> [N, H, W, 5]
             odm_bbox_pred = odm_bbox_pred.transpose([0, 2, 3, 1])
-            odm_bbox_pred_reshape = paddle.reshape(
-                odm_bbox_pred, [-1, 5])
-            odm_bbox_pred_reshape = paddle.unsqueeze(odm_bbox_pred_reshape, axis=0)
+            odm_bbox_pred_reshape = paddle.reshape(odm_bbox_pred, [-1, 5])
+            odm_bbox_pred_reshape = paddle.unsqueeze(
+                odm_bbox_pred_reshape, axis=0)
             odm_reg_branch_list.append(odm_bbox_pred_reshape)
 
         self.s2anet_head_out = (fam_cls_branch_list, fam_reg_branch_list,
@@ -499,12 +501,8 @@ class S2ANetHead(nn.Layer):
         odm_cls_branch_list = self.s2anet_head_out[2]
         odm_reg_branch_list = self.s2anet_head_out[3]
         pred_scores, pred_bboxes = self.get_bboxes(
-            odm_cls_branch_list,
-            odm_reg_branch_list,
-            refine_anchors,
-            nms_pre,
-            self.cls_out_channels,
-            self.use_sigmoid_cls)
+            odm_cls_branch_list, odm_reg_branch_list, refine_anchors, nms_pre,
+            self.cls_out_channels, self.use_sigmoid_cls)
         return pred_scores, pred_bboxes
 
     def smooth_l1_loss(self, pred, label, delta=1.0 / 9.0):
@@ -523,8 +521,8 @@ class S2ANetHead(nn.Layer):
         return loss
 
     def get_fam_loss(self, fam_target, s2anet_head_out, reg_loss_type='gwd'):
-        (labels, label_weights, bbox_targets, bbox_weights, pos_inds,
-         neg_inds) = fam_target
+        (labels, label_weights, bbox_targets, bbox_weights, bbox_gt_bboxes,
+         pos_inds, neg_inds) = fam_target
         fam_cls_branch_list, fam_reg_branch_list, odm_cls_branch_list, odm_reg_branch_list = s2anet_head_out
 
         fam_cls_losses = []
@@ -543,7 +541,6 @@ class S2ANetHead(nn.Layer):
 
             feat_bbox_targets = bbox_targets[st_idx:st_idx + feat_anchor_num, :]
             feat_bbox_weights = bbox_weights[st_idx:st_idx + feat_anchor_num, :]
-            st_idx += feat_anchor_num
 
             # step2: calc cls loss
             feat_labels = feat_labels.reshape(-1)
@@ -590,18 +587,53 @@ class S2ANetHead(nn.Layer):
             fam_bbox_pred = paddle.reshape(fam_bbox_pred, [-1, 5])
             fam_bbox = self.smooth_l1_loss(fam_bbox_pred, feat_bbox_targets)
 
-            # iou_factor
+            fam_bbox = self.smooth_l1_loss(fam_bbox_pred, feat_bbox_targets)
+            loss_weight = paddle.to_tensor(
+                self.reg_loss_weight, dtype='float32', stop_gradient=True)
+            fam_bbox = paddle.multiply(fam_bbox, loss_weight)
+            feat_bbox_weights = paddle.to_tensor(
+                feat_bbox_weights, stop_gradient=True)
+
             if reg_loss_type == 'l1':
-                fam_bbox = self.smooth_l1_loss(fam_bbox_pred, feat_bbox_targets)
-                loss_weight = paddle.to_tensor(
-                    self.reg_loss_weight, dtype='float32', stop_gradient=True)
-                fam_bbox = paddle.multiply(fam_bbox, loss_weight)
-                feat_bbox_weights = paddle.to_tensor(
-                    feat_bbox_weights, stop_gradient=True)
                 fam_bbox = fam_bbox * feat_bbox_weights
                 fam_bbox_total = paddle.sum(fam_bbox) / num_total_samples
+            elif reg_loss_type == 'iou' or reg_loss_type == 'gwd':
+                fam_bbox = paddle.sum(fam_bbox, axis=-1)
+                feat_bbox_weights = paddle.sum(feat_bbox_weights, axis=-1)
+                try:
+                    from rbox_iou_ops import rbox_iou
+                except Exception as e:
+                    print("import custom_ops error, try install rbox_iou_ops " \
+                          "following ppdet/ext_op/README.md", e)
+                    sys.stdout.flush()
+                    sys.exit(-1)
+                # calc iou
+                fam_bbox_decode = self.delta2rbox(self.base_anchors_list[idx],
+                                                  fam_bbox_pred)
+                bbox_gt_bboxes = paddle.to_tensor(
+                    bbox_gt_bboxes,
+                    dtype=fam_bbox_decode.dtype,
+                    place=fam_bbox_decode.place)
+                bbox_gt_bboxes.stop_gradient = True
+                iou = rbox_iou(fam_bbox_decode, bbox_gt_bboxes)
+                iou = paddle.diag(iou)
+
+                if reg_loss_type == 'iou':
+                    EPS = paddle.to_tensor(
+                        1e-8, dtype='float32', stop_gradient=True)
+                    iou_factor = -1.0 * paddle.log(iou + EPS) / (fam_bbox + EPS)
+                    iou_factor.stop_gradient = True
+                    #fam_bbox = fam_bbox * iou_factor
+                elif reg_loss_type == 'gwd':
+                    bbox_gt_bboxes_level = bbox_gt_bboxes[st_idx:st_idx +
+                                                          feat_anchor_num, :]
+                    fam_bbox_total = self.gwd_loss(fam_bbox_decode,
+                                                   bbox_gt_bboxes_level)
+                    fam_bbox_total = fam_bbox_total * feat_bbox_weights
+                    fam_bbox_total = paddle.sum(fam_bbox_total)
 
             fam_bbox_losses.append(fam_bbox_total)
+            st_idx += feat_anchor_num
 
         fam_cls_loss = paddle.add_n(fam_cls_losses)
         fam_cls_loss_weight = paddle.to_tensor(
@@ -611,8 +643,8 @@ class S2ANetHead(nn.Layer):
         return fam_cls_loss, fam_reg_loss
 
     def get_odm_loss(self, odm_target, s2anet_head_out, reg_loss_type='gwd'):
-        (labels, label_weights, bbox_targets, bbox_weights, pos_inds,
-         neg_inds) = odm_target
+        (labels, label_weights, bbox_targets, bbox_weights, bbox_gt_bboxes,
+         pos_inds, neg_inds) = odm_target
         fam_cls_branch_list, fam_reg_branch_list, odm_cls_branch_list, odm_reg_branch_list = s2anet_head_out
 
         odm_cls_losses = []
@@ -621,7 +653,7 @@ class S2ANetHead(nn.Layer):
         num_total_samples = len(pos_inds) + len(
             neg_inds) if self.sampling else len(pos_inds)
         num_total_samples = max(1, num_total_samples)
-        
+
         for idx, feat_size in enumerate(self.featmap_sizes_list):
             feat_anchor_num = feat_size[0] * feat_size[1]
 
@@ -631,7 +663,6 @@ class S2ANetHead(nn.Layer):
 
             feat_bbox_targets = bbox_targets[st_idx:st_idx + feat_anchor_num, :]
             feat_bbox_weights = bbox_weights[st_idx:st_idx + feat_anchor_num, :]
-            st_idx += feat_anchor_num
 
             # step2: calc cls loss
             feat_labels = feat_labels.reshape(-1)
@@ -676,19 +707,53 @@ class S2ANetHead(nn.Layer):
             odm_bbox_pred = paddle.squeeze(odm_bbox_pred, axis=0)
             odm_bbox_pred = paddle.reshape(odm_bbox_pred, [-1, 5])
             odm_bbox = self.smooth_l1_loss(odm_bbox_pred, feat_bbox_targets)
-            
-            # iou_factor odm not use_iou                             
+
+            loss_weight = paddle.to_tensor(
+                self.reg_loss_weight, dtype='float32', stop_gradient=True)
+            odm_bbox = paddle.multiply(odm_bbox, loss_weight)
+            feat_bbox_weights = paddle.to_tensor(
+                feat_bbox_weights, stop_gradient=True)
+
             if reg_loss_type == 'l1':
-                odm_bbox = self.smooth_l1_loss(odm_bbox_pred, feat_bbox_targets)
-                loss_weight = paddle.to_tensor(
-                    self.reg_loss_weight, dtype='float32', stop_gradient=True)
-                odm_bbox = paddle.multiply(odm_bbox, loss_weight)
-                feat_bbox_weights = paddle.to_tensor(
-                    feat_bbox_weights, stop_gradient=True)
                 odm_bbox = odm_bbox * feat_bbox_weights
                 odm_bbox_total = paddle.sum(odm_bbox) / num_total_samples
-            
+            elif reg_loss_type == 'iou' or reg_loss_type == 'gwd':
+                odm_bbox = paddle.sum(odm_bbox, axis=-1)
+                feat_bbox_weights = paddle.sum(feat_bbox_weights, axis=-1)
+                try:
+                    from rbox_iou_ops import rbox_iou
+                except Exception as e:
+                    print("import custom_ops error, try install rbox_iou_ops " \
+                          "following ppdet/ext_op/README.md", e)
+                    sys.stdout.flush()
+                    sys.exit(-1)
+                # calc iou
+                odm_bbox_decode = self.delta2rbox(self.refine_anchor_list[idx],
+                                                  odm_bbox_pred)
+                bbox_gt_bboxes = paddle.to_tensor(
+                    bbox_gt_bboxes,
+                    dtype=odm_bbox_decode.dtype,
+                    place=odm_bbox_decode.place)
+                bbox_gt_bboxes.stop_gradient = True
+                iou = rbox_iou(odm_bbox_decode, bbox_gt_bboxes)
+                iou = paddle.diag(iou)
+
+                if reg_loss_type == 'iou':
+                    EPS = paddle.to_tensor(
+                        1e-8, dtype='float32', stop_gradient=True)
+                    iou_factor = -1.0 * paddle.log(iou + EPS) / (odm_bbox + EPS)
+                    iou_factor.stop_gradient = True
+                    # odm_bbox = odm_bbox * iou_factor
+                elif reg_loss_type == 'gwd':
+                    bbox_gt_bboxes_level = bbox_gt_bboxes[st_idx:st_idx +
+                                                          feat_anchor_num, :]
+                    odm_bbox_total = self.gwd_loss(odm_bbox_decode,
+                                                   bbox_gt_bboxes_level)
+                    odm_bbox_total = odm_bbox_total * feat_bbox_weights
+                    odm_bbox_total = paddle.sum(odm_bbox_total)
+
             odm_bbox_losses.append(odm_bbox_total)
+            st_idx += feat_anchor_num
 
         odm_cls_loss = paddle.add_n(odm_cls_losses)
         odm_cls_loss_weight = paddle.to_tensor(
@@ -737,11 +802,12 @@ class S2ANetHead(nn.Layer):
                 fam_reg_loss_lst.append(im_fam_reg_loss)
 
             # ODM
-            np_refine_anchors_list  = paddle.concat(self.refine_anchor_list).numpy()
+            np_refine_anchors_list = paddle.concat(
+                self.refine_anchor_list).numpy()
             np_refine_anchors_list = np.concatenate(np_refine_anchors_list)
             np_refine_anchors_list = np_refine_anchors_list.reshape(-1, 5)
-            im_odm_target = self.anchor_assign(np_refine_anchors_list, gt_bboxes,
-                                               gt_labels, is_crowd)
+            im_odm_target = self.anchor_assign(np_refine_anchors_list,
+                                               gt_bboxes, gt_labels, is_crowd)
 
             if im_odm_target is not None:
                 im_odm_cls_loss, im_odm_reg_loss = self.get_odm_loss(
@@ -841,7 +907,8 @@ class S2ANetHead(nn.Layer):
         deltas = paddle.reshape(deltas, [-1, 5])
         rrois = paddle.reshape(rrois, [-1, 5])
         # fix dy2st bug denorm_deltas = deltas * self.stds + self.means
-        denorm_deltas = paddle.add(paddle.multiply(deltas, self.stds), self.means)
+        denorm_deltas = paddle.add(
+            paddle.multiply(deltas, self.stds), self.means)
 
         dx = denorm_deltas[:, 0]
         dy = denorm_deltas[:, 1]
@@ -872,9 +939,7 @@ class S2ANetHead(nn.Layer):
         bboxes = paddle.stack([gx, gy, gw, gh, ga], axis=-1)
         return bboxes
 
-    def bbox_decode(self,
-                bbox_preds,
-                anchors):
+    def bbox_decode(self, bbox_preds, anchors):
         """decode bbox from deltas
         Args:
             bbox_preds: [N,H,W,5]
@@ -886,3 +951,109 @@ class S2ANetHead(nn.Layer):
         bbox_delta = paddle.reshape(bbox_preds, [-1, 5])
         bboxes = self.delta2rbox(anchors, bbox_delta)
         return bboxes
+
+    def trace(self, A):
+        tr = paddle.diagonal(A, axis1=-2, axis2=-1)
+        tr = paddle.sum(tr, axis=-1)
+        return tr
+
+    def sqrt_newton_schulz_autograd(self, A, numIters):
+        A_shape = A.shape
+        batchSize = A_shape[0]
+        dim = A_shape[1]
+
+        normA = A * A
+        normA = paddle.sum(normA, axis=1)
+        normA = paddle.sum(normA, axis=1)
+        normA = paddle.sqrt(normA)
+        normA1 = normA.reshape([batchSize, 1, 1])
+        Y = paddle.divide(A, paddle.expand_as(normA1, A))
+        I = paddle.eye(dim, dim).reshape([1, dim, dim])
+        l0 = []
+        for i in range(batchSize):
+            l0.append(I)
+        I = paddle.concat(l0, axis=0)
+        I.stop_gradient = False
+        Z = paddle.eye(dim, dim).reshape([1, dim, dim])
+        l1 = []
+        for i in range(batchSize):
+            l1.append(Z)
+        Z = paddle.concat(l1, axis=0)
+        Z.stop_gradient = False
+
+        for i in range(numIters):
+            T = 0.5 * (3.0 * I - Z.bmm(Y))
+            Y = Y.bmm(T)
+            Z = T.bmm(Z)
+        sA = Y * paddle.sqrt(normA1).reshape([batchSize, 1, 1])
+        sA = paddle.expand_as(sA, A)
+        return sA
+
+    def wasserstein_distance_sigma(sigma1, sigma2):
+        wasserstein_distance_item2 = paddle.matmul(
+            sigma1, sigma1) + paddle.matmul(
+                sigma2, sigma2) - 2 * self.sqrt_newton_schulz_autograd(
+                    paddle.matmul(
+                        paddle.matmul(sigma1, paddle.matmul(sigma2, sigma2)),
+                        sigma1), 10)
+        wasserstein_distance_item2 = self.trace(wasserstein_distance_item2)
+
+        return wasserstein_distance_item2
+
+    def xywhr2xyrs(self, xywhr):
+        xywhr = paddle.reshape(xywhr, [-1, 5])
+        xy = xywhr[:, :2]
+        wh = paddle.clip(xywhr[:, 2:4], min=1e-7, max=1e7)
+        r = xywhr[:, 4]
+        cos_r = paddle.cos(r)
+        sin_r = paddle.sin(r)
+        R = paddle.stack(
+            (cos_r, -sin_r, sin_r, cos_r), axis=-1).reshape([-1, 2, 2])
+        S = 0.5 * paddle.nn.functional.diag_embed(wh)
+        return xy, R, S
+
+    def gwd_loss(self,
+                 pred,
+                 target,
+                 fun='log',
+                 tau=1.0,
+                 alpha=1.0,
+                 normalize=False):
+
+        xy_p, R_p, S_p = self.xywhr2xyrs(pred)
+        xy_t, R_t, S_t = self.xywhr2xyrs(target)
+
+        xy_distance = (xy_p - xy_t).square().sum(axis=-1)
+
+        Sigma_p = R_p.matmul(S_p.square()).matmul(R_p.transpose([0, 2, 1]))
+        Sigma_t = R_t.matmul(S_t.square()).matmul(R_t.transpose([0, 2, 1]))
+
+        whr_distance = paddle.diagonal(
+            S_p, axis1=-2, axis2=-1).square().sum(axis=-1)
+
+        whr_distance = whr_distance + paddle.diagonal(
+            S_t, axis1=-2, axis2=-1).square().sum(axis=-1)
+        _t = Sigma_p.matmul(Sigma_t)
+
+        _t_tr = paddle.diagonal(_t, axis1=-2, axis2=-1).sum(axis=-1)
+        _t_det_sqrt = paddle.diagonal(S_p, axis1=-2, axis2=-1).prod(axis=-1)
+        _t_det_sqrt = _t_det_sqrt * paddle.diagonal(
+            S_t, axis1=-2, axis2=-1).prod(axis=-1)
+        whr_distance = whr_distance + (-2) * (
+            (_t_tr + 2 * _t_det_sqrt).clip(0).sqrt())
+
+        distance = (xy_distance + alpha * alpha * whr_distance).clip(0)
+
+        if normalize:
+            wh_p = pred[..., 2:4].clip(min=1e-7, max=1e7)
+            wh_t = target[..., 2:4].clip(min=1e-7, max=1e7)
+            scale = ((wh_p.log() + wh_t.log()).sum(dim=-1) / 4).exp()
+            distance = distance / scale
+
+        if fun == 'log':
+            distance = paddle.log1p(distance)
+
+        if tau >= 1.0:
+            return 1 - 1 / (tau + distance)
+
+        return distance

ppdet/modeling/proposal_generator/target_layer.py

@@ -451,16 +451,17 @@ class RBoxAssigner(object):
         anchors_num = anchors.shape[0]
         bbox_targets = np.zeros_like(anchors)
         bbox_weights = np.zeros_like(anchors)
+        bbox_gt_bboxes = np.zeros_like(anchors)
         pos_labels = np.ones(anchors_num, dtype=np.int32) * -1
         pos_labels_weights = np.zeros(anchors_num, dtype=np.float32)
 
         pos_sampled_anchors = anchors[pos_inds]
-        #print('ancho target pos_inds', pos_inds, len(pos_inds))
         pos_sampled_gt_boxes = gt_bboxes[anchor_gt_bbox_inds[pos_inds]]
         if len(pos_inds) > 0:
             pos_bbox_targets = self.rbox2delta(pos_sampled_anchors,
                                                pos_sampled_gt_boxes)
             bbox_targets[pos_inds, :] = pos_bbox_targets
+            bbox_gt_bboxes[pos_inds, :] = pos_sampled_gt_boxes
             bbox_weights[pos_inds, :] = 1.0
 
             pos_labels[pos_inds] = labels[pos_inds]
@@ -469,4 +470,4 @@ class RBoxAssigner(object):
         if len(neg_inds) > 0:
             pos_labels_weights[neg_inds] = 1.0
         return (pos_labels, pos_labels_weights, bbox_targets, bbox_weights,
-                pos_inds, neg_inds)
+                bbox_gt_bboxes, pos_inds, neg_inds)