（十八）mmdetection源码解读：forward_train

一、forward_train调用过程

书接上文，在函数_call_impl中，最核心的训练过程，self.forward函数，这个函数是在Module子类中实现的，下面是继承关系：
nn.Module->BaseModule->BaseDetector->TwoStageDetector

class Module:
  def _call_impl(self, *input, **kwargs):
        if torch._C._get_tracing_state():
            result = self._slow_forward(*input, **kwargs)
        else:
            result = self.forward(*input, **kwargs)

在BaseDetector类中forward函数，实际上是调用了forward_train实现的。

class BaseDetector(BaseModule, metaclass=ABCMeta):
    def forward(self, img, img_metas, return_loss=True, **kwargs):
        if torch.onnx.is_in_onnx_export():
            assert len(img_metas) == 1
            return self.onnx_export(img[0], img_metas[0])
        if return_loss:
            return self.forward_train(img, img_metas, **kwargs)
        else:
            return self.forward_test(img, img_metas, **kwargs)

在TwoStageDetector类中forward_train函数，实现了前向传播的整个过程。

class TwoStageDetector(BaseDetector):
    def forward_train(self,
                      img,
                      img_metas,
                      gt_bboxes,
                      gt_labels,
                      gt_bboxes_ignore=None,
                      gt_masks=None,
                      proposals=None,
                      **kwargs):

        x = self.extract_feat(img)

        losses = dict()

        # RPN forward and loss
        if self.with_rpn:
            proposal_cfg = self.train_cfg.get('rpn_proposal',
                                              self.test_cfg.rpn)
            rpn_losses, proposal_list = self.rpn_head.forward_train(
                x,
                img_metas,
                gt_bboxes,
                gt_labels=None,
                gt_bboxes_ignore=gt_bboxes_ignore,
                proposal_cfg=proposal_cfg,
                **kwargs)
            losses.update(rpn_losses)
        else:
            proposal_list = proposals

        roi_losses = self.roi_head.forward_train(x, img_metas, proposal_list,
                                                 gt_bboxes, gt_labels,
                                                 gt_bboxes_ignore, gt_masks,
                                                 **kwargs)
        losses.update(roi_losses)

        return losses

二、forward_train函数详解

上面forward_train 函数实现了如下过程：img->backbone->neck->rpn->roi->losses

2.1、extract_feat

    def extract_feat(self, img):
        """Directly extract features from the backbone+neck."""
        x = self.backbone(img)
        if self.with_neck:
            x = self.neck(x)
        return x

2.2、self.rpn_head.forward_train

    def forward_train(self,
                      x,
                      img_metas,
                      gt_bboxes,
                      gt_labels=None,
                      gt_bboxes_ignore=None,
                      proposal_cfg=None,
                      **kwargs):
        outs = self(x)
        if gt_labels is None:
            loss_inputs = outs + (gt_bboxes, img_metas)
        else:
            loss_inputs = outs + (gt_bboxes, gt_labels, img_metas)
        losses = self.loss(*loss_inputs, gt_bboxes_ignore=gt_bboxes_ignore)
        if proposal_cfg is None:
            return losses
        else:
            proposal_list = self.get_bboxes(
                *outs, img_metas=img_metas, cfg=proposal_cfg)
            return losses, proposal_list

2.3、self.roi_head.forward_train

roi_head在BaseRoIHead的子类中有多种实现，这里简单给出其中一个例子。

@HEADS.register_module()
class StandardRoIHead(BaseRoIHead, BBoxTestMixin, MaskTestMixin):
    def forward_train(self,
                      x,
                      img_metas,
                      proposal_list,
                      gt_bboxes,
                      gt_labels,
                      gt_bboxes_ignore=None,
                      gt_masks=None,
                      **kwargs):
        
        # assign gts and sample proposals
        if self.with_bbox or self.with_mask:
            num_imgs = len(img_metas)
            if gt_bboxes_ignore is None:
                gt_bboxes_ignore = [None for _ in range(num_imgs)]
            sampling_results = []
            for i in range(num_imgs):
                assign_result = self.bbox_assigner.assign(
                    proposal_list[i], gt_bboxes[i], gt_bboxes_ignore[i],
                    gt_labels[i])
                sampling_result = self.bbox_sampler.sample(
                    assign_result,
                    proposal_list[i],
                    gt_bboxes[i],
                    gt_labels[i],
                    feats=[lvl_feat[i][None] for lvl_feat in x])
                sampling_results.append(sampling_result)

        losses = dict()
        # bbox head forward and loss
        if self.with_bbox:
            bbox_results = self._bbox_forward_train(x, sampling_results,
                                                    gt_bboxes, gt_labels,
                                                    img_metas)
            losses.update(bbox_results['loss_bbox'])

        # mask head forward and loss
        if self.with_mask:
            mask_results = self._mask_forward_train(x, sampling_results,
                                                    bbox_results['bbox_feats'],
                                                    gt_masks, img_metas)
            losses.update(mask_results['loss_mask'])

        return losses