mmdetection/mmdet/models/detectors/rtdetr.py

import numpy as np
import math
from typing import  Tuple,Dict
from torch import Tensor, nn
from mmdet.registry import MODELS
from mmdet.utils import OptConfigType,ConfigType
from mmdet.structures import OptSampleList
from .base_detr import DetectionTransformer
from . import RtDetrTransformer

@MODELS.register_module()
class RtDetr(DetectionTransformer):
    def _init_layers(self):
        #we use hybird encoder in the neck and put 
        #simple encoder inside decoder with postional encoding
        self.encoder=None
        self.decoder=RtDetrTransformer(**self.decoder)
        self.positional_encoding=None
    def pre_transformer(
            self,
            mlvl_feats: Tuple[Tensor],
            batch_data_samples: OptSampleList = None) -> None:
        print("pre transformer implemented in transformer")
        return None
    def forward_encoder(self, feat: Tensor, feat_mask: Tensor,
                        feat_pos: Tensor, **kwargs) -> None:
        print("forward_encoder not use in rtdetr")
        return None
    def pre_decoder(self, memory: Tensor, **kwargs) -> None:
        print("pre decoder implemented in rtdetr transformer")
        return None
    def forward_decoder(self,memory: Tensor,
                        **kwargs) -> None:
        print("foward decoder implemented in rtdetr transformer")
        return None
    


    def forward_transformer(self,
                            img_feats: Tuple[Tensor],
                            batch_data_samples: OptSampleList = None) -> Dict:
        """Forward process of Transformer, which includes four steps:
        'pre_transformer' -> 'encoder' -> 'pre_decoder' -> 'decoder'. We
        summarized the parameters flow of the existing DETR-like detector,
        which can be illustrated as follow:

        .. code:: text

                 img_feats from hybridencoder & batch_data_samples
                               |
                               V
                      +-----------------+   |
                      | forward_transformer |   
                      +-----------------+   |
                                |           |
                                V           V
                               head_inputs_dict

        Args:
            img_feats (tuple[Tensor]): Tuple of feature maps from neck. Each
                    feature map has shape (bs, dim, H, W).
            batch_data_samples (list[:obj:`DetDataSample`], optional): The
                batch data samples. It usually includes information such
                as `gt_instance` or `gt_panoptic_seg` or `gt_sem_seg`.
                Defaults to None.

        Returns:
            dict: The dictionary of bbox_head function inputs, which always
            includes the `hidden_states` of the decoder output and may contain
            `references` including the initial and intermediate references.
        """
        # def forward(self,feats,pad_mask=None,gt_meta=None):
        # return (out_bboxes, out_logits, enc_topk_bboxes, enc_topk_logits,
        #         dn_meta)
        out_bboxes,\
        out_logits,\
        enc_topk_bboxes,\
        enc_topk_logits,\
        dn_meta = self.decoder(
            feats=img_feats,
            pad_mask=None,
            gt_meta=batch_data_samples)  
        if self.training:
            head_input_dict=dict(
                            enc_outputs_coord=enc_topk_bboxes,
                            enc_outputs_class=enc_topk_logits,
                            dn_meta=dn_meta\
                            )
        else:
            head_input_dict=dict()
        
        decoder_outputs_dict = dict(
            hidden_states=out_logits, references=out_bboxes)
        head_input_dict.update(decoder_outputs_dict)
        return head_input_dict