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

import torch
import torch.nn as nn
import torch.nn.functional as F
import copy

from .matcher import build_matcher
from utils.misc import sigmoid_focal_loss
from utils.box_ops import box_cxcywh_to_xyxy, generalized_box_iou
from utils.distributed_utils import is_dist_avail_and_initialized, get_world_size


class Criterion(nn.Module):
    """ This class computes the loss for DETR.
    The process happens in two steps:
        1) we compute hungarian assignment between ground truth boxes and the outputs of the model
        2) we supervise each pair of matched ground-truth / prediction (supervise class and box)
    """
    def __init__(self, num_classes, matcher, weight_dict, losses, focal_alpha=0.25):
        """ Create the criterion.
        Parameters:
            num_classes: number of object categories, omitting the special no-object category
            matcher: module able to compute a matching between targets and proposals
            weight_dict: dict containing as key the names of the losses and as values their relative weight.
            eos_coef: relative classification weight applied to the no-object category
            losses: list of all the losses to be applied. See get_loss for list of available losses.
        """
        super().__init__()
        self.num_classes = num_classes
        self.matcher = matcher
        self.weight_dict = weight_dict
        self.losses = losses
        self.focal_alpha = focal_alpha


    def _get_src_permutation_idx(self, indices):
        # permute predictions following indices
        batch_idx = torch.cat([torch.full_like(src, i) for i, (src, _) in enumerate(indices)])
        src_idx = torch.cat([src for (src, _) in indices])
        return batch_idx, src_idx


    def _get_tgt_permutation_idx(self, indices):
        # permute targets following indices
        batch_idx = torch.cat([torch.full_like(tgt, i) for i, (_, tgt) in enumerate(indices)])
        tgt_idx = torch.cat([tgt for (_, tgt) in indices])
        return batch_idx, tgt_idx


    def loss_labels(self, outputs, targets, indices, num_boxes):
        """Classification loss (NLL)
        targets dicts must contain the key "labels" containing a tensor of dim [nb_target_boxes]
        """
        assert 'pred_logits' in outputs
        src_logits = outputs['pred_logits']

        idx = self._get_src_permutation_idx(indices)
        target_classes_o = torch.cat([t["labels"][J] for t, (_, J) in zip(targets, indices)]).to(src_logits.device)
        target_classes = torch.full(src_logits.shape[:2], self.num_classes,
                                    dtype=torch.int64, device=src_logits.device)
        target_classes[idx] = target_classes_o

        target_classes_onehot = torch.zeros([src_logits.shape[0], src_logits.shape[1], src_logits.shape[2] + 1],
                                            dtype=src_logits.dtype, layout=src_logits.layout, device=src_logits.device)
        target_classes_onehot.scatter_(2, target_classes.unsqueeze(-1), 1)

        target_classes_onehot = target_classes_onehot[:, :, :-1]
        loss_cls = sigmoid_focal_loss(src_logits, target_classes_onehot, num_boxes, alpha=self.focal_alpha, gamma=2) * \
                  src_logits.shape[1]
        losses = {'loss_cls': loss_cls}

        return losses


    def loss_boxes(self, outputs, targets, indices, num_boxes):
        """Compute the losses related to the bounding boxes, the L1 regression loss and the GIoU loss
           targets dicts must contain the key "boxes" containing a tensor of dim [nb_target_boxes, 4]
           The target boxes are expected in format (center_x, center_y, w, h), normalized by the image size.
        """
        assert 'pred_boxes' in outputs
        idx = self._get_src_permutation_idx(indices)
        src_boxes = outputs['pred_boxes'][idx]
        target_boxes = torch.cat([t['boxes'][i] for t, (_, i) in zip(targets, indices)], dim=0).to(src_boxes.device)

        loss_bbox = F.l1_loss(src_boxes, target_boxes, reduction='none')

        losses = {}
        losses['loss_bbox'] = loss_bbox.sum() / num_boxes

        loss_giou = 1 - torch.diag(generalized_box_iou(
            box_cxcywh_to_xyxy(src_boxes),
            box_cxcywh_to_xyxy(target_boxes)))
        losses['loss_giou'] = loss_giou.sum() / num_boxes
        return losses


    def get_loss(self, loss, outputs, targets, indices, num_boxes, **kwargs):
        loss_map = {
            'labels': self.loss_labels,
            'boxes': self.loss_boxes,
        }
        assert loss in loss_map, f'do you really want to compute {loss} loss?'
        return loss_map[loss](outputs, targets, indices, num_boxes, **kwargs)


    def forward(self, outputs, targets):
        """ This performs the loss computation.
        Parameters:
             outputs: dict of tensors, see the output specification of the model for the format
             targets: list of dicts, such that len(targets) == batch_size.
                      The expected keys in each dict depends on the losses applied, see each loss' doc
        """
        outputs_without_aux = {k: v for k, v in outputs.items() if k != 'aux_outputs'}

        # Retrieve the matching between the outputs of the last layer and the targets
        indices = self.matcher(outputs_without_aux, targets)

        # Compute the average number of target boxes accross all nodes, for normalization purposes
        num_boxes = sum(len(t["labels"]) for t in targets)
        num_boxes = torch.as_tensor([num_boxes], dtype=torch.float, device=next(iter(outputs.values())).device)
        if is_dist_avail_and_initialized():
            torch.distributed.all_reduce(num_boxes)
        num_boxes = torch.clamp(num_boxes / get_world_size(), min=1).item()

        # Compute all the requested losses
        losses = {}
        for loss in self.losses:
            losses.update(self.get_loss(loss, outputs, targets, indices, num_boxes))

        # In case of auxiliary losses, we repeat this process with the output of each intermediate layer.
        if 'aux_outputs' in outputs:
            for i, aux_outputs in enumerate(outputs['aux_outputs']):
                indices = self.matcher(aux_outputs, targets)
                for loss in self.losses:
                    kwargs = {}
                    l_dict = self.get_loss(loss, aux_outputs, targets, indices, num_boxes, **kwargs)
                    l_dict = {k + f'_{i}': v for k, v in l_dict.items()}
                    losses.update(l_dict)

        weight_dict = self.weight_dict
        total_loss = sum(losses[k] * weight_dict[k] for k in losses.keys() if k in weight_dict)
        losses['losses'] = total_loss

        return losses


# build criterion
def build_criterion(cfg, num_classes, aux_loss=False):
    matcher = build_matcher(cfg)
    
    weight_dict = {'loss_cls': cfg['loss_cls_weight'],
                  'loss_bbox': cfg['loss_box_weight'],
                  'loss_giou': cfg['loss_giou_weight']}

    # TODO this is a hack
    if aux_loss:
        aux_weight_dict = {}
        for i in range(cfg['num_decoder_layers'] - 1):
            aux_weight_dict.update({k + f'_{i}': v for k, v in weight_dict.items()})
        weight_dict.update(aux_weight_dict)

    losses = ['labels', 'boxes']
    
    criterion = Criterion(
        num_classes=num_classes,
        matcher=matcher,
        weight_dict=weight_dict,
        losses=losses,
        focal_alpha=cfg['focal_alpha'])

    return criterion