use AlignedSimOTA & QFL for RTCDet-v2

config/model_config/rtcdet_v2_config.py

@@ -38,25 +38,24 @@ rtcdet_v2_cfg = {
         'num_cls_head': 2,
         'num_reg_head': 2,
         'head_depthwise': False,
-        'reg_max': 16,
         # ---------------- Train config ----------------
         ## Input
         'multi_scale': [0.5, 1.25],   # 320 -> 800
         'trans_type': 'yolox_small',
         # ---------------- Assignment config ----------------
         ## Matcher
-        'matcher': {'ota': {'center_sampling_radius': 2.5,
-                             'topk_candidate': 10},
+        'matcher': {'soft_center_radius': 3.0,
+                    'topk_candidate': 13,
+                    'iou_weight': 3.0
                     },
         # ---------------- Loss config ----------------
         ## Loss weight
         'ema_update': False,
         'loss_box_aux': True,
         'loss_cls_weight': 1.0,
-        'loss_box_weight': 5.0,
-        'loss_dfl_weight': 1.0,
+        'loss_box_weight': 2.0,
         # ---------------- Train config ----------------
-        'trainer_type': 'yolox',
+        'trainer_type': 'rtmdet',
     },
 
     'rtcdet_v2_l':{
@@ -95,25 +94,24 @@ rtcdet_v2_cfg = {
         'num_cls_head': 2,
         'num_reg_head': 2,
         'head_depthwise': False,
-        'reg_max': 16,
         # ---------------- Train config ----------------
         ## Input
         'multi_scale': [0.5, 1.25],   # 320 -> 800
         'trans_type': 'yolox_large',
         # ---------------- Assignment config ----------------
         ## Matcher
-        'matcher': {'ota': {'center_sampling_radius': 2.5,
-                             'topk_candidate': 10},
+        'matcher': {'soft_center_radius': 3.0,
+                    'topk_candidate': 13,
+                    'iou_weight': 3.0
                     },
         # ---------------- Loss config ----------------
         ## Loss weight
         'ema_update': False,
         'loss_box_aux': True,
         'loss_cls_weight': 1.0,
-        'loss_box_weight': 5.0,
-        'loss_dfl_weight': 1.0,
+        'loss_box_weight': 2.0,
         # ---------------- Train config ----------------
-        'trainer_type': 'yolox',
+        'trainer_type': 'rtmdet',
     },
 
 }

models/detectors/rtcdet_v2/loss.py

@@ -19,15 +19,15 @@ class Criterion(object):
         # ---------------- Loss weight ----------------
         self.loss_cls_weight = cfg['loss_cls_weight']
         self.loss_box_weight = cfg['loss_box_weight']
-        self.loss_dfl_weight = cfg['loss_dfl_weight']
         self.loss_box_aux    = cfg['loss_box_aux']
         # ---------------- Matcher ----------------
         matcher_config = cfg['matcher']
-        ## SimOTA assigner
+        ## Aligned SimOTA assigner
         self.ota_matcher = AlignedSimOTA(
-            center_sampling_radius=matcher_config['ota']['center_sampling_radius'],
-            topk_candidate=matcher_config['ota']['topk_candidate'],
-            num_classes=num_classes
+            num_classes=num_classes,
+            soft_center_radius=matcher_config['soft_center_radius'],
+            topk_candidate=matcher_config['topk_candidate'],
+            iou_weight=matcher_config['iou_weight']
         )
 
 
@@ -41,76 +41,56 @@ class Criterion(object):
         return new
 
 
-    def loss_classes(self, pred_cls, gt_score, gt_label=None, vfl=False):
-        if vfl:
-            assert gt_label is not None
-            # compute varifocal loss
-            alpha, gamma = 0.75, 2.0
-            focal_weight = alpha * pred_cls.sigmoid().pow(gamma) * (1 - gt_label) + gt_score * gt_label
-            bce_loss = F.binary_cross_entropy_with_logits(pred_cls, gt_score, reduction='none')
-            loss_cls = bce_loss * focal_weight
-        else:
-            # compute bce loss
-            loss_cls = F.binary_cross_entropy_with_logits(pred_cls, gt_score, reduction='none')
+    def loss_classes(self, pred_cls, target, beta=2.0):
+        # Quality FocalLoss
+        """
+            pred_cls: (torch.Tensor): [N, C]。
+            target:   (tuple([torch.Tensor], [torch.Tensor])): label -> [N,], score -> [N,]
+        """
+        label, score = target
+        pred_sigmoid = pred_cls.sigmoid()
+        scale_factor = pred_sigmoid
+        zerolabel = scale_factor.new_zeros(pred_cls.shape)
+
+        ce_loss = F.binary_cross_entropy_with_logits(
+            pred_cls, zerolabel, reduction='none') * scale_factor.pow(beta)
+        
+        bg_class_ind = pred_cls.shape[-1]
+        pos = ((label >= 0) & (label < bg_class_ind)).nonzero().squeeze(1)
+        pos_label = label[pos].long()
 
-        return loss_cls
+        scale_factor = score[pos] - pred_sigmoid[pos, pos_label]
 
+        ce_loss[pos, pos_label] = F.binary_cross_entropy_with_logits(
+            pred_cls[pos, pos_label], score[pos],
+            reduction='none') * scale_factor.abs().pow(beta)
 
-    def loss_bboxes(self, pred_box, gt_box, bbox_weight=None):
+        return ce_loss
+    
+
+    def loss_bboxes(self, pred_box, gt_box):
         # regression loss
         ious = get_ious(pred_box, gt_box, 'xyxy', 'giou')
         loss_box = 1.0 - ious
 
-        if bbox_weight is not None:
-            loss_box *= bbox_weight
-
         return loss_box
 
 
-    def loss_dfl(self, pred_reg, gt_box, anchor, stride, bbox_weight=None):
-        # rescale coords by stride
-        gt_box_s = gt_box / stride
-        anchor_s = anchor / stride
-
-        # compute deltas
-        gt_ltrb_s = bbox2dist(anchor_s, gt_box_s, self.cfg['reg_max'] - 1)
-
-        gt_left = gt_ltrb_s.to(torch.long)
-        gt_right = gt_left + 1
-
-        weight_left = gt_right.to(torch.float) - gt_ltrb_s
-        weight_right = 1 - weight_left
-
-        # loss left
-        loss_left = F.cross_entropy(
-            pred_reg.view(-1, self.cfg['reg_max']),
-            gt_left.view(-1),
-            reduction='none').view(gt_left.shape) * weight_left
-        # loss right
-        loss_right = F.cross_entropy(
-            pred_reg.view(-1, self.cfg['reg_max']),
-            gt_right.view(-1),
-            reduction='none').view(gt_left.shape) * weight_right
-
-        loss_dfl = (loss_left + loss_right).mean(-1)
-        
-        if bbox_weight is not None:
-            loss_dfl *= bbox_weight
-
-        return loss_dfl
-
-
-    def loss_bboxes_aux(self, pred_delta, gt_box, anchors, stride_tensors):
-        gt_delta_tl = (anchors - gt_box[..., :2]) / stride_tensors
-        gt_delta_rb = (gt_box[..., 2:] - anchors) / stride_tensors
-        gt_delta = torch.cat([gt_delta_tl, gt_delta_rb], dim=1)
-        loss_box_aux = F.l1_loss(pred_delta, gt_delta, reduction='none')
+    def loss_bboxes_aux(self, pred_reg, gt_box, anchors, stride_tensors):
+        # xyxy -> cxcy&bwbh
+        gt_cxcy = (gt_box[..., :2] + gt_box[..., 2:]) * 0.5
+        gt_bwbh = gt_box[..., 2:] - gt_box[..., :2]
+        # encode gt box
+        gt_cxcy_encode = (gt_cxcy - anchors) / stride_tensors
+        gt_bwbh_encode = torch.log(gt_bwbh / stride_tensors)
+        gt_box_encode = torch.cat([gt_cxcy_encode, gt_bwbh_encode], dim=-1)
+        # l1 loss
+        loss_box_aux = F.l1_loss(pred_reg, gt_box_encode, reduction='none')
 
         return loss_box_aux
 
 
     def __call__(self, outputs, targets, epoch=0):
-        """ Compute loss with SimOTA assigner """
         bs = outputs['pred_cls'][0].shape[0]
         device = outputs['pred_cls'][0].device
         fpn_strides = outputs['strides']
@@ -124,48 +104,31 @@ class Criterion(object):
         # --------------- label assignment ---------------
         cls_targets = []
         box_targets = []
-        fg_masks = []
+        assign_metrics = []
         for batch_idx in range(bs):
-            tgt_labels = targets[batch_idx]["labels"].to(device)
-            tgt_bboxes = targets[batch_idx]["boxes"].to(device)
-
-            # check target
-            if len(tgt_labels) == 0 or tgt_bboxes.max().item() == 0.:
-                # There is no valid gt
-                cls_target = cls_preds.new_zeros((num_anchors, self.num_classes))
-                box_target = cls_preds.new_zeros((0, 4))
-                fg_mask = cls_preds.new_zeros(num_anchors).bool()
-            else:
-                (
-                    fg_mask,
-                    assigned_labels,
-                    assigned_ious,
-                    assigned_indexs
-                ) = self.ota_matcher(
-                    fpn_strides = fpn_strides,
-                    anchors = anchors,
-                    pred_cls = cls_preds[batch_idx], 
-                    pred_box = box_preds[batch_idx],
-                    tgt_labels = tgt_labels,
-                    tgt_bboxes = tgt_bboxes
-                    )
-                # prepare cls targets
-                assigned_labels = F.one_hot(assigned_labels.long(), self.num_classes)
-                assigned_labels = assigned_labels * assigned_ious.unsqueeze(-1)
-                cls_target = assigned_labels.new_zeros((num_anchors, self.num_classes))
-                cls_target[fg_mask] = assigned_labels
-                # prepare box targets
-                box_target = tgt_bboxes[assigned_indexs]
-
-            cls_targets.append(cls_target)
-            box_targets.append(box_target)
-            fg_masks.append(fg_mask)
-
-        cls_targets = torch.cat(cls_targets, 0)
-        box_targets = torch.cat(box_targets, 0)
-        fg_masks = torch.cat(fg_masks, 0)
-        num_fgs = fg_masks.sum()
-
+            tgt_labels = targets[batch_idx]["labels"].to(device)  # [N,]
+            tgt_bboxes = targets[batch_idx]["boxes"].to(device)   # [N, 4]
+            # label assignment
+            assigned_result = self.ota_matcher(fpn_strides=fpn_strides,
+                                               anchors=anchors,
+                                               pred_cls=cls_preds[batch_idx].detach(),
+                                               pred_box=box_preds[batch_idx].detach(),
+                                               gt_labels=tgt_labels,
+                                               gt_bboxes=tgt_bboxes
+                                              )
+            cls_targets.append(assigned_result['assigned_labels'])
+            box_targets.append(assigned_result['assigned_bboxes'])
+            assign_metrics.append(assigned_result['assign_metrics'])
+        
+        cls_targets = torch.cat(cls_targets, dim=0)
+        box_targets = torch.cat(box_targets, dim=0)
+        assign_metrics = torch.cat(assign_metrics, dim=0)
+
+        # FG cat_id: [0, num_classes -1], BG cat_id: num_classes
+        bg_class_ind = self.num_classes
+        pos_inds = ((cls_targets >= 0) & (cls_targets < bg_class_ind)).nonzero().squeeze(1)
+        num_fgs = assign_metrics.sum()
+        
         # average loss normalizer across all the GPUs
         if is_dist_avail_and_initialized():
             torch.distributed.all_reduce(num_fgs)
@@ -179,49 +142,38 @@ class Criterion(object):
         
         # ------------------ Classification loss ------------------
         cls_preds = cls_preds.view(-1, self.num_classes)
-        loss_cls = self.loss_classes(cls_preds, cls_targets)
+        loss_cls = self.loss_classes(cls_preds, (cls_targets, assign_metrics))
         loss_cls = loss_cls.sum() / normalizer
 
         # ------------------ Regression loss ------------------
-        box_preds_pos = box_preds.view(-1, 4)[fg_masks]
-        loss_box = self.loss_bboxes(box_preds_pos, box_targets)
+        box_preds_pos = box_preds.view(-1, 4)[pos_inds]
+        box_targets_pos = box_targets[pos_inds]
+        loss_box = self.loss_bboxes(box_preds_pos, box_targets_pos)
         loss_box = loss_box.sum() / normalizer
 
-        # ------------------ Distribution focal loss  ------------------
-        ## process anchors
-        anchors = torch.cat(anchors, dim=0)
-        anchors = anchors[None].repeat(bs, 1, 1).view(-1, 2)
-        ## process stride tensors
-        strides = torch.cat(outputs['stride_tensor'], dim=0)
-        strides = strides.unsqueeze(0).repeat(bs, 1, 1).view(-1, 1)
-        ## fg preds
-        reg_preds_pos = reg_preds.view(-1, 4*self.cfg['reg_max'])[fg_masks]
-        anchors_pos = anchors[fg_masks]
-        strides_pos = strides[fg_masks]
-        ## compute dfl
-        loss_dfl = self.loss_dfl(reg_preds_pos, box_targets, anchors_pos, strides_pos)
-        loss_dfl = loss_dfl.sum() / normalizer
-
-        # total loss
         losses = self.loss_cls_weight * loss_cls + \
-                 self.loss_box_weight * loss_box + \
-                 self.loss_dfl_weight * loss_dfl
+                 self.loss_box_weight * loss_box
 
         loss_dict = dict(
                 loss_cls = loss_cls,
                 loss_box = loss_box,
-                loss_dfl = loss_dfl,
                 losses = losses
         )
 
         # ------------------ Aux regression loss ------------------
-        if epoch >= (self.max_epoch - self.no_aug_epoch - 1) and self.loss_box_aux:
-            ## delta_preds
-            delta_preds = torch.cat(outputs['pred_delta'], dim=1)
-            delta_preds_pos = delta_preds.view(-1, 4)[fg_masks]
+        if epoch >= (self.max_epoch - self.no_aug_epoch - 1):
+            ## reg_preds
+            reg_preds = torch.cat(outputs['pred_reg'], dim=1)
+            reg_preds_pos = reg_preds.view(-1, 4)[pos_inds]
+            ## anchor tensors
+            anchors_tensors = torch.cat(outputs['anchors'], dim=0)[None].repeat(bs, 1, 1)
+            anchors_tensors_pos = anchors_tensors.view(-1, 2)[pos_inds]
+            ## stride tensors
+            stride_tensors = torch.cat(outputs['stride_tensors'], dim=0)[None].repeat(bs, 1, 1)
+            stride_tensors_pos = stride_tensors.view(-1, 1)[pos_inds]
             ## aux loss
-            loss_box_aux = self.loss_bboxes_aux(delta_preds_pos, box_targets, anchors_pos, strides_pos)
-            loss_box_aux = loss_box_aux.sum() / num_fgs
+            loss_box_aux = self.loss_bboxes_aux(reg_preds_pos, box_targets_pos, anchors_tensors_pos, stride_tensors_pos)
+            loss_box_aux = loss_box_aux.sum() / normalizer
 
             losses += loss_box_aux
             loss_dict['loss_box_aux'] = loss_box_aux

models/detectors/rtcdet_v2/matcher.py

@@ -1,210 +1,23 @@
-import torch
-import torch.nn as nn
-import torch.nn.functional as F
-from utils.box_ops import box_iou, bbox_iou
-
-
-# -------------------------- Basic Functions --------------------------
-def select_candidates_in_gts(xy_centers, gt_bboxes, eps=1e-9):
-    """select the positive anchors's center in gt
-    Args:
-        xy_centers (Tensor): shape(bs*n_max_boxes, num_total_anchors, 4)
-        gt_bboxes (Tensor): shape(bs, n_max_boxes, 4)
-    Return:
-        (Tensor): shape(bs, n_max_boxes, num_total_anchors)
-    """
-    n_anchors = xy_centers.size(0)
-    bs, n_max_boxes, _ = gt_bboxes.size()
-    _gt_bboxes = gt_bboxes.reshape([-1, 4])
-    xy_centers = xy_centers.unsqueeze(0).repeat(bs * n_max_boxes, 1, 1)
-    gt_bboxes_lt = _gt_bboxes[:, 0:2].unsqueeze(1).repeat(1, n_anchors, 1)
-    gt_bboxes_rb = _gt_bboxes[:, 2:4].unsqueeze(1).repeat(1, n_anchors, 1)
-    b_lt = xy_centers - gt_bboxes_lt
-    b_rb = gt_bboxes_rb - xy_centers
-    bbox_deltas = torch.cat([b_lt, b_rb], dim=-1)
-    bbox_deltas = bbox_deltas.reshape([bs, n_max_boxes, n_anchors, -1])
-    return (bbox_deltas.min(axis=-1)[0] > eps).to(gt_bboxes.dtype)
-
-def select_highest_overlaps(mask_pos, overlaps, n_max_boxes):
-    """if an anchor box is assigned to multiple gts,
-        the one with the highest iou will be selected.
-    Args:
-        mask_pos (Tensor): shape(bs, n_max_boxes, num_total_anchors)
-        overlaps (Tensor): shape(bs, n_max_boxes, num_total_anchors)
-    Return:
-        target_gt_idx (Tensor): shape(bs, num_total_anchors)
-        fg_mask (Tensor): shape(bs, num_total_anchors)
-        mask_pos (Tensor): shape(bs, n_max_boxes, num_total_anchors)
-    """
-    fg_mask = mask_pos.sum(axis=-2)
-    if fg_mask.max() > 1:
-        mask_multi_gts = (fg_mask.unsqueeze(1) > 1).repeat([1, n_max_boxes, 1])
-        max_overlaps_idx = overlaps.argmax(axis=1)
-        is_max_overlaps = F.one_hot(max_overlaps_idx, n_max_boxes)
-        is_max_overlaps = is_max_overlaps.permute(0, 2, 1).to(overlaps.dtype)
-        mask_pos = torch.where(mask_multi_gts, is_max_overlaps, mask_pos)
-        fg_mask = mask_pos.sum(axis=-2)
-    target_gt_idx = mask_pos.argmax(axis=-2)
-    return target_gt_idx, fg_mask , mask_pos
-
-def iou_calculator(box1, box2, eps=1e-9):
-    """Calculate iou for batch
-    Args:
-        box1 (Tensor): shape(bs, n_max_boxes, 1, 4)
-        box2 (Tensor): shape(bs, 1, num_total_anchors, 4)
-    Return:
-        (Tensor): shape(bs, n_max_boxes, num_total_anchors)
-    """
-    box1 = box1.unsqueeze(2)  # [N, M1, 4] -> [N, M1, 1, 4]
-    box2 = box2.unsqueeze(1)  # [N, M2, 4] -> [N, 1, M2, 4]
-    px1y1, px2y2 = box1[:, :, :, 0:2], box1[:, :, :, 2:4]
-    gx1y1, gx2y2 = box2[:, :, :, 0:2], box2[:, :, :, 2:4]
-    x1y1 = torch.maximum(px1y1, gx1y1)
-    x2y2 = torch.minimum(px2y2, gx2y2)
-    overlap = (x2y2 - x1y1).clip(0).prod(-1)
-    area1 = (px2y2 - px1y1).clip(0).prod(-1)
-    area2 = (gx2y2 - gx1y1).clip(0).prod(-1)
-    union = area1 + area2 - overlap + eps
-
-    return overlap / union
-
-
-# -------------------------- Task Aligned Assigner --------------------------
-class TaskAlignedAssigner(nn.Module):
-    def __init__(self, topk=10, alpha=0.5, beta=6.0, eps=1e-9, num_classes=80):
-        super(TaskAlignedAssigner, self).__init__()
-        self.topk = topk
-        self.num_classes = num_classes
-        self.bg_idx = num_classes
-        self.alpha = alpha
-        self.beta = beta
-        self.eps = eps
-
-    @torch.no_grad()
-    def forward(self,
-                pd_scores,
-                pd_bboxes,
-                anc_points,
-                gt_labels,
-                gt_bboxes):
-        """This code referenced to
-           https://github.com/Nioolek/PPYOLOE_pytorch/blob/master/ppyoloe/assigner/tal_assigner.py
-        Args:
-            pd_scores (Tensor): shape(bs, num_total_anchors, num_classes)
-            pd_bboxes (Tensor): shape(bs, num_total_anchors, 4)
-            anc_points (Tensor): shape(num_total_anchors, 2)
-            gt_labels (Tensor): shape(bs, n_max_boxes, 1)
-            gt_bboxes (Tensor): shape(bs, n_max_boxes, 4)
-        Returns:
-            target_labels (Tensor): shape(bs, num_total_anchors)
-            target_bboxes (Tensor): shape(bs, num_total_anchors, 4)
-            target_scores (Tensor): shape(bs, num_total_anchors, num_classes)
-            fg_mask (Tensor): shape(bs, num_total_anchors)
-        """
-        self.bs = pd_scores.size(0)
-        self.n_max_boxes = gt_bboxes.size(1)
-
-        mask_pos, align_metric, overlaps = self.get_pos_mask(
-            pd_scores, pd_bboxes, gt_labels, gt_bboxes, anc_points)
-
-        target_gt_idx, fg_mask, mask_pos = select_highest_overlaps(
-            mask_pos, overlaps, self.n_max_boxes)
-
-        # assigned target
-        target_labels, target_bboxes, target_scores = self.get_targets(
-            gt_labels, gt_bboxes, target_gt_idx, fg_mask)
-
-        # normalize
-        align_metric *= mask_pos
-        pos_align_metrics = align_metric.amax(axis=-1, keepdim=True)  # b, max_num_obj
-        pos_overlaps = (overlaps * mask_pos).amax(axis=-1, keepdim=True)  # b, max_num_obj
-        norm_align_metric = (align_metric * pos_overlaps / (pos_align_metrics + self.eps)).amax(-2).unsqueeze(-1)
-        target_scores = target_scores * norm_align_metric
-
-        return target_labels, target_bboxes, target_scores, fg_mask.bool(), target_gt_idx
-
-
-    def get_pos_mask(self, pd_scores, pd_bboxes, gt_labels, gt_bboxes, anc_points):
-        # get anchor_align metric, (b, max_num_obj, h*w)
-        align_metric, overlaps = self.get_box_metrics(pd_scores, pd_bboxes, gt_labels, gt_bboxes)
-        # get in_gts mask, (b, max_num_obj, h*w)
-        mask_in_gts = select_candidates_in_gts(anc_points, gt_bboxes)
-        # get topk_metric mask, (b, max_num_obj, h*w)
-        mask_topk = self.select_topk_candidates(align_metric * mask_in_gts)
-        # merge all mask to a final mask, (b, max_num_obj, h*w)
-        mask_pos = mask_topk * mask_in_gts
-
-        return mask_pos, align_metric, overlaps
-
-
-    def get_box_metrics(self, pd_scores, pd_bboxes, gt_labels, gt_bboxes):
-        ind = torch.zeros([2, self.bs, self.n_max_boxes], dtype=torch.long)  # 2, b, max_num_obj
-        ind[0] = torch.arange(end=self.bs).view(-1, 1).repeat(1, self.n_max_boxes)  # b, max_num_obj
-        ind[1] = gt_labels.long().squeeze(-1)  # b, max_num_obj
-        # get the scores of each grid for each gt cls
-        bbox_scores = pd_scores[ind[0], :, ind[1]]  # b, max_num_obj, h*w
-
-        overlaps = bbox_iou(gt_bboxes.unsqueeze(2), pd_bboxes.unsqueeze(1), xywh=False).squeeze(3).clamp(0)
-        align_metric = bbox_scores.pow(self.alpha) * overlaps.pow(self.beta)
-
-        return align_metric, overlaps
-
-
-    def select_topk_candidates(self, metrics, largest=True):
-        """
-        Args:
-            metrics: (b, max_num_obj, h*w).
-            topk_mask: (b, max_num_obj, topk) or None
-        """
+# ---------------------------------------------------------------------
+# Copyright (c) OpenMMLab. All rights reserved.
+# ---------------------------------------------------------------------
 
-        num_anchors = metrics.shape[-1]  # h*w
-        # (b, max_num_obj, topk)
-        topk_metrics, topk_idxs = torch.topk(metrics, self.topk, dim=-1, largest=largest)
-        topk_mask = (topk_metrics.max(-1, keepdim=True)[0] > self.eps).tile([1, 1, self.topk])
-        # (b, max_num_obj, topk)
-        topk_idxs[~topk_mask] = 0
-        # (b, max_num_obj, topk, h*w) -> (b, max_num_obj, h*w)
-        is_in_topk = F.one_hot(topk_idxs, num_anchors).sum(-2)
-        # filter invalid bboxes
-        is_in_topk = torch.where(is_in_topk > 1, 0, is_in_topk)
-        return is_in_topk.to(metrics.dtype)
 
+import torch
+import torch.nn.functional as F
+from utils.box_ops import *
 
-    def get_targets(self, gt_labels, gt_bboxes, target_gt_idx, fg_mask):
-        """
-        Args:
-            gt_labels: (b, max_num_obj, 1)
-            gt_bboxes: (b, max_num_obj, 4)
-            target_gt_idx: (b, h*w)
-            fg_mask: (b, h*w)
-        """
-
-        # assigned target labels, (b, 1)
-        batch_ind = torch.arange(end=self.bs, dtype=torch.int64, device=gt_labels.device)[..., None]
-        target_gt_idx = target_gt_idx + batch_ind * self.n_max_boxes  # (b, h*w)
-        target_labels = gt_labels.long().flatten()[target_gt_idx]  # (b, h*w)
-
-        # assigned target boxes, (b, max_num_obj, 4) -> (b, h*w)
-        target_bboxes = gt_bboxes.view(-1, 4)[target_gt_idx]
-
-        # assigned target scores
-        target_labels.clamp(0)
-        target_scores = F.one_hot(target_labels, self.num_classes)  # (b, h*w, 80)
-        fg_scores_mask = fg_mask[:, :, None].repeat(1, 1, self.num_classes)  # (b, h*w, 80)
-        target_scores = torch.where(fg_scores_mask > 0, target_scores, 0)
-
-        return target_labels, target_bboxes, target_scores
-    
 
-# -------------------------- Aligned SimOTA Assigner --------------------------
+# RTMDet's Assigner
 class AlignedSimOTA(object):
     """
-        This code referenced to https://github.com/Megvii-BaseDetection/YOLOX/blob/main/yolox/models/yolo_head.py
+        This code referenced to https://github.com/open-mmlab/mmyolo/models/task_modules/assigners/batch_dsl_assigner.py
     """
-    def __init__(self, num_classes, center_sampling_radius, topk_candidate ):
+    def __init__(self, num_classes=80, soft_center_radius=3.0, topk_candidate=13, iou_weight=3.0):
         self.num_classes = num_classes
-        self.center_sampling_radius = center_sampling_radius
+        self.soft_center_radius = soft_center_radius
         self.topk_candidate = topk_candidate
+        self.iou_weight = iou_weight
 
 
     @torch.no_grad()
@@ -213,161 +26,151 @@ class AlignedSimOTA(object):
                  anchors, 
                  pred_cls, 
                  pred_box, 
-                 tgt_labels,
-                 tgt_bboxes):
+                 gt_labels,
+                 gt_bboxes):
         # [M,]
-        strides_tensor = torch.cat([torch.ones_like(anchor_i[:, 0]) * stride_i
+        strides = torch.cat([torch.ones_like(anchor_i[:, 0]) * stride_i
                                 for stride_i, anchor_i in zip(fpn_strides, anchors)], dim=-1)
         # List[F, M, 2] -> [M, 2]
         anchors = torch.cat(anchors, dim=0)
-        num_anchor = anchors.shape[0]        
-        num_gt = len(tgt_labels)
-
-        # ----------------------- Find inside points -----------------------
-        fg_mask, is_in_boxes_and_center = self.get_in_boxes_info(
-            tgt_bboxes, anchors, strides_tensor, num_anchor, num_gt)
-        cls_preds = pred_cls[fg_mask].float()   # [Mp, C]
-        box_preds = pred_box[fg_mask].float()   # [Mp, 4]
-
-        # ----------------------- Reg cost -----------------------
-        pair_wise_ious, _ = box_iou(tgt_bboxes, box_preds)      # [N, Mp]
-        reg_cost = -torch.log(pair_wise_ious + 1e-8)            # [N, Mp]
-
-        # ----------------------- Cls cost -----------------------
-        with torch.cuda.amp.autocast(enabled=False):
-            # [Mp, C] -> [N, Mp, C]
-            score_preds = cls_preds.sigmoid_().unsqueeze(0).repeat(num_gt, 1, 1)
-            # prepare cls_target
-            cls_targets = F.one_hot(tgt_labels.long(), self.num_classes).float()
-            cls_targets = cls_targets.unsqueeze(1).repeat(1, score_preds.size(1), 1)
-            cls_targets *= pair_wise_ious.unsqueeze(-1)  # iou-aware
-            # [N, Mp]
-            cls_cost = F.binary_cross_entropy(score_preds, cls_targets, reduction="none").sum(-1)
-        del score_preds
-
-        #----------------------- Dynamic K-Matching -----------------------
-        cost_matrix = (
-            cls_cost
-            + 3.0 * reg_cost
-            + 100000.0 * (~is_in_boxes_and_center)
-        ) # [N, Mp]
-
+        num_gt = len(gt_labels)
+
+        # check gt
+        if num_gt == 0 or gt_bboxes.max().item() == 0.:
+            return {
+                'assigned_labels': gt_labels.new_full(pred_cls[..., 0].shape,
+                                                      self.num_classes,
+                                                      dtype=torch.long),
+                'assigned_bboxes': gt_bboxes.new_full(pred_box.shape, 0),
+                'assign_metrics': gt_bboxes.new_full(pred_cls[..., 0].shape, 0)
+            }
+        
+        # get inside points: [N, M]
+        is_in_gt = self.find_inside_points(gt_bboxes, anchors)
+        valid_mask = is_in_gt.sum(dim=0) > 0  # [M,]
+
+        # ----------------------------------- soft center prior -----------------------------------
+        gt_center = (gt_bboxes[..., :2] + gt_bboxes[..., 2:]) / 2.0
+        distance = (anchors.unsqueeze(0) - gt_center.unsqueeze(1)
+                    ).pow(2).sum(-1).sqrt() / strides.unsqueeze(0)  # [N, M]
+        distance = distance * valid_mask.unsqueeze(0)
+        soft_center_prior = torch.pow(10, distance - self.soft_center_radius)
+
+        # ----------------------------------- regression cost -----------------------------------
+        pair_wise_ious, _ = box_iou(gt_bboxes, pred_box)  # [N, M]
+        pair_wise_ious_loss = -torch.log(pair_wise_ious + 1e-8) * self.iou_weight
+
+        # ----------------------------------- classification cost -----------------------------------
+        ## select the predicted scores corresponded to the gt_labels
+        pairwise_pred_scores = pred_cls.permute(1, 0)  # [M, C] -> [C, M]
+        pairwise_pred_scores = pairwise_pred_scores[gt_labels.long(), :].float()   # [N, M]
+        ## scale factor
+        scale_factor = (pair_wise_ious - pairwise_pred_scores.sigmoid()).abs().pow(2.0)
+        ## cls cost
+        pair_wise_cls_loss = F.binary_cross_entropy_with_logits(
+            pairwise_pred_scores, pair_wise_ious,
+            reduction="none") * scale_factor # [N, M]
+            
+        del pairwise_pred_scores
+
+        ## foreground cost matrix
+        cost_matrix = pair_wise_cls_loss + pair_wise_ious_loss + soft_center_prior
+        max_pad_value = torch.ones_like(cost_matrix) * 1e9
+        cost_matrix = torch.where(valid_mask[None].repeat(num_gt, 1),   # [N, M]
+                                  cost_matrix, max_pad_value)
+
+        # ----------------------------------- dynamic label assignment -----------------------------------
         (
-            assigned_labels,         # [num_fg,]
-            assigned_ious,           # [num_fg,]
-            assigned_indexs,         # [num_fg,]
+            matched_pred_ious,
+            matched_gt_inds,
+            fg_mask_inboxes
         ) = self.dynamic_k_matching(
             cost_matrix,
             pair_wise_ious,
-            tgt_labels,
-            num_gt,
-            fg_mask
+            num_gt
             )
-        del cls_cost, cost_matrix, pair_wise_ious, reg_cost
-
-        return fg_mask, assigned_labels, assigned_ious, assigned_indexs
+        del pair_wise_cls_loss, cost_matrix, pair_wise_ious, pair_wise_ious_loss
 
+        # -----------------------------------process assigned labels -----------------------------------
+        assigned_labels = gt_labels.new_full(pred_cls[..., 0].shape,
+                                             self.num_classes)  # [M,]
+        assigned_labels[fg_mask_inboxes] = gt_labels[matched_gt_inds].squeeze(-1)
+        assigned_labels = assigned_labels.long()  # [M,]
 
-    def get_in_boxes_info(
-        self,
-        gt_bboxes,   # [N, 4]
-        anchors,     # [M, 2]
-        strides,     # [M,]
-        num_anchors, # M
-        num_gt,      # N
-        ):
-        # anchor center
-        x_centers = anchors[:, 0]
-        y_centers = anchors[:, 1]
+        assigned_bboxes = gt_bboxes.new_full(pred_box.shape, 0)        # [M, 4]
+        assigned_bboxes[fg_mask_inboxes] = gt_bboxes[matched_gt_inds]  # [M, 4]
 
-        # [M,] -> [1, M] -> [N, M]
-        x_centers = x_centers.unsqueeze(0).repeat(num_gt, 1)
-        y_centers = y_centers.unsqueeze(0).repeat(num_gt, 1)
+        assign_metrics = gt_bboxes.new_full(pred_cls[..., 0].shape, 0) # [M, 4]
+        assign_metrics[fg_mask_inboxes] = matched_pred_ious            # [M, 4]
 
-        # [N,] -> [N, 1] -> [N, M]
-        gt_bboxes_l = gt_bboxes[:, 0].unsqueeze(1).repeat(1, num_anchors) # x1
-        gt_bboxes_t = gt_bboxes[:, 1].unsqueeze(1).repeat(1, num_anchors) # y1
-        gt_bboxes_r = gt_bboxes[:, 2].unsqueeze(1).repeat(1, num_anchors) # x2
-        gt_bboxes_b = gt_bboxes[:, 3].unsqueeze(1).repeat(1, num_anchors) # y2
-
-        b_l = x_centers - gt_bboxes_l
-        b_r = gt_bboxes_r - x_centers
-        b_t = y_centers - gt_bboxes_t
-        b_b = gt_bboxes_b - y_centers
-        bbox_deltas = torch.stack([b_l, b_t, b_r, b_b], 2)
+        assigned_dict = dict(
+            assigned_labels=assigned_labels,
+            assigned_bboxes=assigned_bboxes,
+            assign_metrics=assign_metrics
+            )
+        
+        return assigned_dict
 
-        is_in_boxes = bbox_deltas.min(dim=-1).values > 0.0
-        is_in_boxes_all = is_in_boxes.sum(dim=0) > 0
-        # in fixed center
-        center_radius = self.center_sampling_radius
 
-        # [N, 2]
-        gt_centers = (gt_bboxes[:, :2] + gt_bboxes[:, 2:]) * 0.5
-        
-        # [1, M]
-        center_radius_ = center_radius * strides.unsqueeze(0)
+    def find_inside_points(self, gt_bboxes, anchors):
+        """
+            gt_bboxes: Tensor -> [N, 2]
+            anchors:   Tensor -> [M, 2]
+        """
+        num_anchors = anchors.shape[0]
+        num_gt = gt_bboxes.shape[0]
 
-        gt_bboxes_l = gt_centers[:, 0].unsqueeze(1).repeat(1, num_anchors) - center_radius_ # x1
-        gt_bboxes_t = gt_centers[:, 1].unsqueeze(1).repeat(1, num_anchors) - center_radius_ # y1
-        gt_bboxes_r = gt_centers[:, 0].unsqueeze(1).repeat(1, num_anchors) + center_radius_ # x2
-        gt_bboxes_b = gt_centers[:, 1].unsqueeze(1).repeat(1, num_anchors) + center_radius_ # y2
+        anchors_expand = anchors.unsqueeze(0).repeat(num_gt, 1, 1)           # [N, M, 2]
+        gt_bboxes_expand = gt_bboxes.unsqueeze(1).repeat(1, num_anchors, 1)  # [N, M, 4]
 
-        c_l = x_centers - gt_bboxes_l
-        c_r = gt_bboxes_r - x_centers
-        c_t = y_centers - gt_bboxes_t
-        c_b = gt_bboxes_b - y_centers
-        center_deltas = torch.stack([c_l, c_t, c_r, c_b], 2)
-        is_in_centers = center_deltas.min(dim=-1).values > 0.0
-        is_in_centers_all = is_in_centers.sum(dim=0) > 0
+        # offset
+        lt = anchors_expand - gt_bboxes_expand[..., :2]
+        rb = gt_bboxes_expand[..., 2:] - anchors_expand
+        bbox_deltas = torch.cat([lt, rb], dim=-1)
 
-        # in boxes and in centers
-        is_in_boxes_anchor = is_in_boxes_all | is_in_centers_all
+        is_in_gts = bbox_deltas.min(dim=-1).values > 0
 
-        is_in_boxes_and_center = (
-            is_in_boxes[:, is_in_boxes_anchor] & is_in_centers[:, is_in_boxes_anchor]
-        )
-        return is_in_boxes_anchor, is_in_boxes_and_center
-    
+        return is_in_gts
     
-    def dynamic_k_matching(
-        self, 
-        cost, 
-        pair_wise_ious, 
-        gt_classes, 
-        num_gt, 
-        fg_mask
-        ):
-        # Dynamic K
-        # ---------------------------------------------------------------
-        matching_matrix = torch.zeros_like(cost, dtype=torch.uint8)
 
-        ious_in_boxes_matrix = pair_wise_ious
-        n_candidate_k = min(self.topk_candidate, ious_in_boxes_matrix.size(1))
-        topk_ious, _ = torch.topk(ious_in_boxes_matrix, n_candidate_k, dim=1)
+    def dynamic_k_matching(self, cost_matrix, pairwise_ious, num_gt):
+        """Use IoU and matching cost to calculate the dynamic top-k positive
+        targets.
+
+        Args:
+            cost_matrix (Tensor): Cost matrix.
+            pairwise_ious (Tensor): Pairwise iou matrix.
+            num_gt (int): Number of gt.
+            valid_mask (Tensor): Mask for valid bboxes.
+        Returns:
+            tuple: matched ious and gt indexes.
+        """
+        matching_matrix = torch.zeros_like(cost_matrix, dtype=torch.uint8)
+        # select candidate topk ious for dynamic-k calculation
+        candidate_topk = min(self.topk_candidate, pairwise_ious.size(1))
+        topk_ious, _ = torch.topk(pairwise_ious, candidate_topk, dim=1)
+        # calculate dynamic k for each gt
         dynamic_ks = torch.clamp(topk_ious.sum(1).int(), min=1)
-        dynamic_ks = dynamic_ks.tolist()
-        for gt_idx in range(num_gt):
-            _, pos_idx = torch.topk(
-                cost[gt_idx], k=dynamic_ks[gt_idx], largest=False
-            )
-            matching_matrix[gt_idx][pos_idx] = 1
 
-        del topk_ious, dynamic_ks, pos_idx
+        # sorting the batch cost matirx is faster than topk
+        _, sorted_indices = torch.sort(cost_matrix, dim=1)
+        for gt_idx in range(num_gt):
+            topk_ids = sorted_indices[gt_idx, :dynamic_ks[gt_idx]]
+            matching_matrix[gt_idx, :][topk_ids] = 1
 
-        anchor_matching_gt = matching_matrix.sum(0)
-        if (anchor_matching_gt > 1).sum() > 0:
-            _, cost_argmin = torch.min(cost[:, anchor_matching_gt > 1], dim=0)
-            matching_matrix[:, anchor_matching_gt > 1] *= 0
-            matching_matrix[cost_argmin, anchor_matching_gt > 1] = 1
-        fg_mask_inboxes = matching_matrix.sum(0) > 0
+        del topk_ious, dynamic_ks, topk_ids
 
-        fg_mask[fg_mask.clone()] = fg_mask_inboxes
+        prior_match_gt_mask = matching_matrix.sum(0) > 1
+        if prior_match_gt_mask.sum() > 0:
+            cost_min, cost_argmin = torch.min(
+                cost_matrix[:, prior_match_gt_mask], dim=0)
+            matching_matrix[:, prior_match_gt_mask] *= 0
+            matching_matrix[cost_argmin, prior_match_gt_mask] = 1
 
-        assigned_indexs = matching_matrix[:, fg_mask_inboxes].argmax(0)
-        assigned_labels = gt_classes[assigned_indexs]
+        # get foreground mask inside box and center prior
+        fg_mask_inboxes = matching_matrix.sum(0) > 0
+        matched_pred_ious = (matching_matrix *
+                             pairwise_ious).sum(0)[fg_mask_inboxes]
+        matched_gt_inds = matching_matrix[:, fg_mask_inboxes].argmax(0)
 
-        assigned_ious = (matching_matrix * pair_wise_ious).sum(0)[
-            fg_mask_inboxes
-        ]
-        return assigned_labels, assigned_ious, assigned_indexs
-    
+        return matched_pred_ious, matched_gt_inds, fg_mask_inboxes

models/detectors/rtcdet_v2/rtcdet_v2.py

@@ -18,18 +18,17 @@ class RTCDet(nn.Module):
     def __init__(self, 
                  cfg,
                  device, 
-                 num_classes = 20, 
-                 conf_thresh = 0.05,
-                 nms_thresh = 0.6,
-                 trainable = False, 
-                 topk = 1000,
-                 deploy = False):
+                 num_classes :int   = 20, 
+                 conf_thresh :float = 0.05,
+                 nms_thresh  :float = 0.6,
+                 topk        :int   = 1000,
+                 trainable   :bool  = False, 
+                 deploy      :bool  = False):
         super(RTCDet, self).__init__()
         # ---------------------- Basic Parameters ----------------------
         self.cfg = cfg
         self.device = device
         self.stride = cfg['stride']
-        self.reg_max = cfg['reg_max']
         self.num_classes = num_classes
         self.trainable = trainable
         self.conf_thresh = conf_thresh
@@ -52,7 +51,7 @@ class RTCDet(nn.Module):
 
         ## ----------- Heads -----------
         self.det_heads = build_det_head(
-            cfg, self.fpn_dims, self.head_dim, num_classes, self.reg_max, num_levels=len(self.stride))
+            cfg, self.fpn_dims, self.head_dim, num_classes, num_levels=len(self.stride))
 
         ## ----------- Preds -----------
         self.pred_layers = build_pred_layer(

models/detectors/rtcdet_v2/rtcdet_v2_head.py

@@ -72,7 +72,7 @@ class SingleLevelHead(nn.Module):
 
 # Multi-level Head
 class MultiLevelHead(nn.Module):
-    def __init__(self, cfg, in_dims, out_dim, num_classes=80, reg_max=16, num_levels=3):
+    def __init__(self, cfg, in_dims, out_dim, num_classes=80, num_levels=3):
         super().__init__()
         ## ----------- Network Parameters -----------
         self.multi_level_heads = nn.ModuleList(
@@ -112,9 +112,9 @@ class MultiLevelHead(nn.Module):
     
 
 # build detection head
-def build_det_head(cfg, in_dim, out_dim, num_classes=80, reg_max=16, num_levels=3):
+def build_det_head(cfg, in_dim, out_dim, num_classes=80, num_levels=3):
     if cfg['head'] == 'decoupled_head':
-        head = MultiLevelHead(cfg, in_dim, out_dim, num_classes, reg_max, num_levels) 
+        head = MultiLevelHead(cfg, in_dim, out_dim, num_classes, num_levels) 
 
     return head
 

models/detectors/rtcdet_v2/rtcdet_v2_pred.py

@@ -49,7 +49,7 @@ class SingleLevelPredLayer(nn.Module):
 
 # Multi-level pred layer
 class MultiLevelPredLayer(nn.Module):
-    def __init__(self, cls_dim, reg_dim, strides, num_classes, num_coords=4, num_levels=3, reg_max=16):
+    def __init__(self, cls_dim, reg_dim, strides, num_classes, num_coords=4, num_levels=3):
         super().__init__()
         # --------- Basic Parameters ----------
         self.cls_dim = cls_dim
@@ -58,7 +58,6 @@ class MultiLevelPredLayer(nn.Module):
         self.num_classes = num_classes
         self.num_coords = num_coords
         self.num_levels = num_levels
-        self.reg_max = reg_max
 
         # ----------- Network Parameters -----------
         ## pred layers
@@ -67,13 +66,9 @@ class MultiLevelPredLayer(nn.Module):
                 cls_dim,
                 reg_dim,
                 num_classes,
-                num_coords * self.reg_max)
+                num_coords)
                 for _ in range(num_levels)
             ])
-        ## proj conv
-        self.proj = nn.Parameter(torch.linspace(0, reg_max, reg_max), requires_grad=False)
-        self.proj_conv = nn.Conv2d(self.reg_max, 1, kernel_size=1, bias=False)
-        self.proj_conv.weight = nn.Parameter(self.proj.view([1, reg_max, 1, 1]).clone().detach(), requires_grad=False)
 
 
     def generate_anchors(self, level, fmp_size):
@@ -97,7 +92,6 @@ class MultiLevelPredLayer(nn.Module):
         all_cls_preds = []
         all_reg_preds = []
         all_box_preds = []
-        all_delta_preds = []
         for level in range(self.num_levels):
             # pred
             cls_pred, reg_pred = self.multi_level_preds[level](cls_feats[level], reg_feats[level])
@@ -112,27 +106,18 @@ class MultiLevelPredLayer(nn.Module):
             
             # [B, C, H, W] -> [B, H, W, C] -> [B, M, C]
             cls_pred = cls_pred.permute(0, 2, 3, 1).contiguous().view(B, -1, self.num_classes)
-            reg_pred = reg_pred.permute(0, 2, 3, 1).contiguous().view(B, -1, 4*self.reg_max)
+            reg_pred = reg_pred.permute(0, 2, 3, 1).contiguous().view(B, -1, 4)
 
             # ----------------------- Decode bbox -----------------------
-            B, M = reg_pred.shape[:2]
-            # [B, M, 4*(reg_max)] -> [B, M, 4, reg_max] -> [B, 4, M, reg_max]
-            delta_pred = reg_pred.reshape([B, M, 4, self.reg_max])
-            # [B, M, 4, reg_max] -> [B, reg_max, 4, M]
-            delta_pred = delta_pred.permute(0, 3, 2, 1).contiguous()
-            # [B, reg_max, 4, M] -> [B, 1, 4, M]
-            delta_pred = self.proj_conv(F.softmax(delta_pred, dim=1))
-            # [B, 1, 4, M] -> [B, 4, M] -> [B, M, 4]
-            delta_pred = delta_pred.view(B, 4, M).permute(0, 2, 1).contiguous()
-            ## tlbr -> xyxy
-            x1y1_pred = anchors[None] - delta_pred[..., :2] * self.strides[level]
-            x2y2_pred = anchors[None] + delta_pred[..., 2:] * self.strides[level]
-            box_pred = torch.cat([x1y1_pred, x2y2_pred], dim=-1)
+            ctr_pred = reg_pred[..., :2] * self.strides[level] + anchors[..., :2]
+            wh_pred = torch.exp(reg_pred[..., 2:]) * self.strides[level]
+            pred_x1y1 = ctr_pred - wh_pred * 0.5
+            pred_x2y2 = ctr_pred + wh_pred * 0.5
+            box_pred = torch.cat([pred_x1y1, pred_x2y2], dim=-1)
 
             all_cls_preds.append(cls_pred)
             all_reg_preds.append(reg_pred)
             all_box_preds.append(box_pred)
-            all_delta_preds.append(delta_pred)
             all_anchors.append(anchors)
             all_strides.append(stride_tensor)
         
@@ -140,10 +125,9 @@ class MultiLevelPredLayer(nn.Module):
         outputs = {"pred_cls": all_cls_preds,        # List(Tensor) [B, M, C]
                    "pred_reg": all_reg_preds,        # List(Tensor) [B, M, 4*(reg_max)]
                    "pred_box": all_box_preds,        # List(Tensor) [B, M, 4]
-                   "pred_delta": all_delta_preds,    # List(Tensor) [B, M, 4]
                    "anchors": all_anchors,           # List(Tensor) [M, 2]
                    "strides": self.strides,          # List(Int) = [8, 16, 32]
-                   "stride_tensor": all_strides      # List(Tensor) [M, 1]
+                   "stride_tensors": all_strides      # List(Tensor) [M, 1]
                    }
 
         return outputs

train_multi_gpus.sh

@@ -4,8 +4,8 @@ python -m torch.distributed.run --nproc_per_node=8 train.py \
                                                     -dist \
                                                     -d coco \
                                                     --root /data/datasets/ \
-                                                    -m yolox_s\
-                                                    -bs 64 \
+                                                    -m yolov7_tiny\
+                                                    -bs 128 \
                                                     -size 640 \
                                                     --wp_epoch 3 \
                                                     --max_epoch 300 \