YOLO-Tutorial-v2/yolo/models/yolov10/yolov10.py

# --------------- Torch components ---------------
import copy
import torch
import torch.nn as nn

# --------------- Model components ---------------
from .yolov10_backbone import Yolov10Backbone
from .yolov10_pafpn    import Yolov10PaFPN
from .yolov10_head     import Yolov10DetHead
from .yolov10_pred     import Yolov10DetPredLayer


# YOLOv10
class Yolov10(nn.Module):
    def __init__(self,
                 cfg,
                 is_val = False,
                 ) -> None:
        super(Yolov10, self).__init__()
        # ---------------------- Basic setting ----------------------
        self.cfg = cfg
        self.num_classes = cfg.num_classes
        ## Post-process parameters
        self.topk_candidates  = cfg.val_topk        if is_val else cfg.test_topk
        self.conf_thresh      = cfg.val_conf_thresh if is_val else cfg.test_conf_thresh
        self.nms_thresh       = cfg.val_nms_thresh  if is_val else cfg.test_nms_thresh
        self.no_multi_labels  = False if is_val else True
        
        # ---------------------- Network Parameters ----------------------
        ## Backbone
        self.backbone = Yolov10Backbone(cfg)
        self.pyramid_feat_dims = self.backbone.feat_dims[-3:]

        ## PaFPN
        self.fpn = Yolov10PaFPN(cfg, self.backbone.feat_dims)

        ## Head
        self.head_o2m = Yolov10DetHead(cfg, self.fpn.out_dims)
        self.pred_o2m = Yolov10DetPredLayer(cfg, self.head_o2m.cls_head_dim, self.head_o2m.reg_head_dim)

        self.head_o2o = copy.deepcopy(self.head_o2m)
        self.pred_o2o = copy.deepcopy(self.pred_o2m)

    def post_process(self, cls_preds, box_preds):
        """
        We process predictions at each scale hierarchically
        Input:
            cls_preds: List[torch.Tensor] -> [[B, M, C], ...], B=1
            box_preds: List[torch.Tensor] -> [[B, M, 4], ...], B=1
        Output:
            bboxes: np.array -> [N, 4]
            scores: np.array -> [N,]
            labels: np.array -> [N,]
        """
        all_scores = []
        all_labels = []
        all_bboxes = []
        
        for cls_pred_i, box_pred_i in zip(cls_preds, box_preds):
            cls_pred_i = cls_pred_i[0]
            box_pred_i = box_pred_i[0]
            if self.no_multi_labels:
                # [M,]
                scores, labels = torch.max(cls_pred_i.sigmoid(), dim=1)

                # Keep top k top scoring indices only.
                num_topk = min(self.topk_candidates, box_pred_i.size(0))

                # topk candidates
                predicted_prob, topk_idxs = scores.sort(descending=True)
                topk_scores = predicted_prob[:num_topk]
                topk_idxs = topk_idxs[:num_topk]

                # filter out the proposals with low confidence score
                keep_idxs = topk_scores > self.conf_thresh
                scores = topk_scores[keep_idxs]
                topk_idxs = topk_idxs[keep_idxs]

                labels = labels[topk_idxs]
                bboxes = box_pred_i[topk_idxs]
            else:
                # [M, C] -> [MC,]
                scores_i = cls_pred_i.sigmoid().flatten()

                # Keep top k top scoring indices only.
                num_topk = min(self.topk_candidates, box_pred_i.size(0))

                # torch.sort is actually faster than .topk (at least on GPUs)
                predicted_prob, topk_idxs = scores_i.sort(descending=True)
                topk_scores = predicted_prob[:num_topk]
                topk_idxs = topk_idxs[:num_topk]

                # filter out the proposals with low confidence score
                keep_idxs = topk_scores > self.conf_thresh
                scores = topk_scores[keep_idxs]
                topk_idxs = topk_idxs[keep_idxs]

                anchor_idxs = torch.div(topk_idxs, self.num_classes, rounding_mode='floor')
                labels = topk_idxs % self.num_classes

                bboxes = box_pred_i[anchor_idxs]

            all_scores.append(scores)
            all_labels.append(labels)
            all_bboxes.append(bboxes)

        scores = torch.cat(all_scores, dim=0)
        labels = torch.cat(all_labels, dim=0)
        bboxes = torch.cat(all_bboxes, dim=0)

        # to cpu & numpy
        scores = scores.cpu().numpy()
        labels = labels.cpu().numpy()
        bboxes = bboxes.cpu().numpy()
        
        # keep top-300 results
        scores = scores[:300]
        bboxes = bboxes[:300]
        labels = labels[:300]
        
        return bboxes, scores, labels
    
    def forward(self, x):
        # ---------------- Backbone ----------------
        pyramid_feats = self.backbone(x)

        # ---------------- PaFPN ----------------
        pyramid_feats = self.fpn(pyramid_feats)

        # ---------------- Heads (one-to-one) ----------------
        pyramid_feats_detach = [feat.detach() for feat in pyramid_feats]
        cls_feats, reg_feats = self.head_o2o(pyramid_feats_detach)
        outputs_o2o = self.pred_o2o(cls_feats, reg_feats)
        outputs_o2o['image_size'] = [x.shape[2], x.shape[3]]

        if not self.training:
            all_cls_preds = outputs_o2o['pred_cls']
            all_box_preds = outputs_o2o['pred_box']

            # post process
            bboxes, scores, labels = self.post_process(all_cls_preds, all_box_preds)
            outputs = {
                "scores": scores,
                "labels": labels,
                "bboxes": bboxes
            }
        else:
            # ---------------- Heads (one-to-many) ----------------
            cls_feats, reg_feats = self.head_o2m(pyramid_feats)
            outputs_o2m = self.pred_o2m(cls_feats, reg_feats)
            outputs_o2m['image_size'] = [x.shape[2], x.shape[3]]

            outputs = {
                "outputs_o2o": outputs_o2o,
                "outputs_o2m": outputs_o2m,
            }
            

        return outputs