RT-ODLab/models/detectors/yolovx/loss.py

import torch
import torch.nn.functional as F

from utils.box_ops import  bbox2dist, get_ious
from utils.distributed_utils import get_world_size, is_dist_avail_and_initialized

from .matcher import AlignedSimOTA


class Criterion(object):
    def __init__(self, 
                 cfg, 
                 device, 
                 num_classes=80):
        self.cfg = cfg
        self.device = device
        self.num_classes = num_classes
        # loss weight
        self.loss_obj_weight = cfg['loss_obj_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']
        # matcher
        matcher_config = cfg['matcher']
        self.matcher = AlignedSimOTA(
            num_classes=num_classes,
            center_sampling_radius=matcher_config['center_sampling_radius'],
            topk_candidate=matcher_config['topk_candicate']
            )


    def loss_objectness(self, pred_obj, gt_obj):
        loss_obj = F.binary_cross_entropy_with_logits(pred_obj, gt_obj, reduction='none')

        return loss_obj
    

    def loss_classes(self, pred_cls, gt_label):
        loss_cls = F.binary_cross_entropy_with_logits(pred_cls, gt_label, reduction='none')

        return loss_cls


    def loss_bboxes(self, pred_box, gt_box):
        # regression loss
        ious = get_ious(pred_box, gt_box, "xyxy", 'giou')
        loss_box = 1.0 - ious

        return loss_box


    def loss_dfl(self, pred_reg, gt_box, anchor, stride):
        # 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, keepdim=True)
            
        return loss_dfl
    
    
    def __call__(self, outputs, targets, epoch=0):        
        """
            outputs['pred_obj']: List(Tensor) [B, M, 1]
            outputs['pred_cls']: List(Tensor) [B, M, C]
            outputs['pred_box']: List(Tensor) [B, M, 4]
            outputs['strides']: List(Int) [8, 16, 32] output stride
            targets: (List) [dict{'boxes': [...], 
                                 'labels': [...], 
                                 'orig_size': ...}, ...]
        """
        bs = outputs['pred_cls'][0].shape[0]
        device = outputs['pred_cls'][0].device
        fpn_strides = outputs['strides']
        anchors = outputs['anchors']
        # preds: [B, M, C]
        obj_preds = torch.cat(outputs['pred_obj'], dim=1)
        cls_preds = torch.cat(outputs['pred_cls'], dim=1)
        reg_preds = torch.cat(outputs['pred_reg'], dim=1)
        box_preds = torch.cat(outputs['pred_box'], dim=1)

        # --------------- label assignment ---------------
        obj_targets = []
        cls_targets = []
        box_targets = []
        fg_masks = []
        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.:
                num_anchors = sum([ab.shape[0] for ab in anchors])
                # There is no valid gt
                cls_target = obj_preds.new_zeros((0, self.num_classes))
                box_target = obj_preds.new_zeros((0, 4))
                obj_target = obj_preds.new_zeros((num_anchors, 1))
                fg_mask = obj_preds.new_zeros(num_anchors).bool()
            else:
                (
                    fg_mask,
                    assigned_labels,
                    assigned_ious,
                    assigned_indexs
                ) = self.matcher(
                    fpn_strides = fpn_strides,
                    anchors = anchors,
                    pred_obj = obj_preds[batch_idx],
                    pred_cls = cls_preds[batch_idx], 
                    pred_box = box_preds[batch_idx],
                    tgt_labels = tgt_labels,
                    tgt_bboxes = tgt_bboxes
                    )

                obj_target = fg_mask.unsqueeze(-1)
                cls_target = F.one_hot(assigned_labels.long(), self.num_classes)
                cls_target = cls_target * assigned_ious.unsqueeze(-1)
                box_target = tgt_bboxes[assigned_indexs]

            cls_targets.append(cls_target)
            box_targets.append(box_target)
            obj_targets.append(obj_target)
            fg_masks.append(fg_mask)

        cls_targets = torch.cat(cls_targets, 0)
        box_targets = torch.cat(box_targets, 0)
        obj_targets = torch.cat(obj_targets, 0)
        fg_masks = torch.cat(fg_masks, 0)
        num_fgs = fg_masks.sum()

        if is_dist_avail_and_initialized():
            torch.distributed.all_reduce(num_fgs)
        num_fgs = (num_fgs / get_world_size()).clamp(1.0)

        # ------------------ Objecntness loss ------------------
        loss_obj = self.loss_objectness(obj_preds.view(-1, 1), obj_targets.float())
        loss_obj = loss_obj.sum() / num_fgs
        
        # ------------------ Classification loss ------------------
        cls_preds_pos = cls_preds.view(-1, self.num_classes)[fg_masks]
        loss_cls = self.loss_classes(cls_preds_pos, cls_targets)
        loss_cls = loss_cls.sum() / num_fgs

        # ------------------ Regression loss ------------------
        box_preds_pos = box_preds.view(-1, 4)[fg_masks]
        loss_box = self.loss_bboxes(box_preds_pos, box_targets)
        loss_box = loss_box.sum() / num_fgs

        # ------------------ 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() / num_fgs

        # total loss
        losses = self.loss_obj_weight * loss_obj + \
                 self.loss_cls_weight * loss_cls + \
                 self.loss_box_weight * loss_box + \
                 self.loss_dfl_weight * loss_dfl

        loss_dict = dict(
                loss_obj = loss_obj,
                loss_cls = loss_cls,
                loss_box = loss_box,
                loss_dfl = loss_dfl,
                losses = losses
        )

        return loss_dict
    

def build_criterion(cfg, device, num_classes):
    criterion = Criterion(
        cfg=cfg,
        device=device,
        num_classes=num_classes
        )

    return criterion


if __name__ == "__main__":
    pass