simplify non batch nms (open-mmlab#99)

mmdeploy/codebase/mmdet/core/post_processing/bbox_nms.py

@@ -5,6 +5,7 @@
 import mmdeploy
 from mmdeploy.core import FUNCTION_REWRITER, mark
 from mmdeploy.mmcv.ops import ONNXNMSop, TRTBatchedNMSop
+from mmdeploy.utils import is_dynamic_batch
 
 
 def select_nms_index(scores: torch.Tensor,
@@ -82,28 +83,10 @@ def _multiclass_nms(boxes: Tensor,
                     keep_top_k: int = -1):
     """Create a dummy onnx::NonMaxSuppression op while exporting to ONNX.
 
-    This function helps exporting to onnx with batch and multiclass NMS op.
-    It only supports class-agnostic detection results. That is, the scores
-    is of shape (N, num_bboxes, num_classes) and the boxes is of shape
-    (N, num_boxes, 4).
-
-    Args:
-        boxes (Tensor): The bounding boxes of shape [N, num_boxes, 4].
-        scores (Tensor): The detection scores of shape
-            [N, num_boxes, num_classes].
-        max_output_boxes_per_class (int): Maximum number of output
-            boxes per class of nms. Defaults to 1000.
-        iou_threshold (float): IOU threshold of nms. Defaults to 0.5.
-        score_threshold (float): score threshold of nms.
-            Defaults to 0.05.
-        pre_top_k (int): Number of top K boxes to keep before nms.
-            Defaults to -1.
-        keep_top_k (int): Number of top K boxes to keep after nms.
-            Defaults to -1.
-
-    Returns:
-        tuple[Tensor, Tensor]: (dets, labels), `dets` of shape [N, num_det, 5]
-            and `labels` of shape [N, num_det].
+    This function helps exporting to onnx with batch and multiclass NMS op. It
+    only supports class-agnostic detection results. That is, the scores is of
+    shape (N, num_bboxes, num_classes) and the boxes is of shape (N, num_boxes,
+    4).
     """
     max_output_boxes_per_class = torch.LongTensor([max_output_boxes_per_class])
     iou_threshold = torch.tensor([iou_threshold], dtype=torch.float32)
@@ -129,6 +112,116 @@ def _multiclass_nms(boxes: Tensor,
     return dets, labels
 
 
+def _multiclass_nms_single(boxes: Tensor,
+                           scores: Tensor,
+                           max_output_boxes_per_class: int = 1000,
+                           iou_threshold: float = 0.5,
+                           score_threshold: float = 0.05,
+                           pre_top_k: int = -1,
+                           keep_top_k: int = -1):
+    """Create a dummy onnx::NonMaxSuppression op while exporting to ONNX.
+
+    Single batch nms could be optimized.
+    """
+    max_output_boxes_per_class = torch.LongTensor([max_output_boxes_per_class])
+    iou_threshold = torch.tensor([iou_threshold], dtype=torch.float32)
+    score_threshold = torch.tensor([score_threshold], dtype=torch.float32)
+
+    # pre topk
+    if pre_top_k > 0:
+        max_scores, _ = scores.max(-1)
+        _, topk_inds = max_scores.squeeze(0).topk(pre_top_k)
+        boxes = boxes[:, topk_inds, :]
+        scores = scores[:, topk_inds, :]
+
+    scores = scores.permute(0, 2, 1)
+    selected_indices = ONNXNMSop.apply(boxes, scores,
+                                       max_output_boxes_per_class,
+                                       iou_threshold, score_threshold)
+
+    cls_inds = selected_indices[:, 1]
+    box_inds = selected_indices[:, 2]
+
+    scores = scores[:, cls_inds, box_inds].unsqueeze(2)
+    boxes = boxes[:, box_inds, ...]
+    dets = torch.cat([boxes, scores], dim=2)
+    labels = cls_inds.unsqueeze(0)
+
+    # pad
+    dets = torch.cat((dets, dets.new_zeros((1, 1, 5))), 1)
+    labels = torch.cat((labels, labels.new_zeros((1, 1))), 1)
+
+    # topk or sort
+    is_use_topk = keep_top_k > 0 and \
+        (torch.onnx.is_in_onnx_export() or keep_top_k < dets.shape[1])
+    if is_use_topk:
+        _, topk_inds = dets[:, :, -1].topk(keep_top_k, dim=1)
+    else:
+        _, topk_inds = dets[:, :, -1].sort(dim=1, descending=True)
+    topk_inds = topk_inds.squeeze(0)
+    dets = dets[:, topk_inds, ...]
+    labels = labels[:, topk_inds, ...]
+
+    return dets, labels
+
+
+@FUNCTION_REWRITER.register_rewriter(
+    func_name='mmdeploy.codebase.mmdet.core.post_processing._multiclass_nms')
+def multiclass_nms__default(ctx,
+                            boxes: Tensor,
+                            scores: Tensor,
+                            max_output_boxes_per_class: int = 1000,
+                            iou_threshold: float = 0.5,
+                            score_threshold: float = 0.05,
+                            pre_top_k: int = -1,
+                            keep_top_k: int = -1):
+    """Create a dummy onnx::NonMaxSuppression op while exporting to ONNX.
+
+    This function helps exporting to onnx with batch and multiclass NMS op.
+    It only supports class-agnostic detection results. That is, the scores
+    is of shape (N, num_bboxes, num_classes) and the boxes is of shape
+    (N, num_boxes, 4).
+
+    Args:
+        boxes (Tensor): The bounding boxes of shape [N, num_boxes, 4].
+        scores (Tensor): The detection scores of shape
+            [N, num_boxes, num_classes].
+        max_output_boxes_per_class (int): Maximum number of output
+            boxes per class of nms. Defaults to 1000.
+        iou_threshold (float): IOU threshold of nms. Defaults to 0.5.
+        score_threshold (float): score threshold of nms.
+            Defaults to 0.05.
+        pre_top_k (int): Number of top K boxes to keep before nms.
+            Defaults to -1.
+        keep_top_k (int): Number of top K boxes to keep after nms.
+            Defaults to -1.
+
+    Returns:
+        tuple[Tensor, Tensor]: (dets, labels), `dets` of shape [N, num_det, 5]
+            and `labels` of shape [N, num_det].
+    """
+    deploy_cfg = ctx.cfg
+    batch_size = boxes.size(0)
+    if not is_dynamic_batch(deploy_cfg) and batch_size != 1:
+        return _multiclass_nms_single(
+            boxes,
+            scores,
+            max_output_boxes_per_class=max_output_boxes_per_class,
+            iou_threshold=iou_threshold,
+            score_threshold=score_threshold,
+            pre_top_k=pre_top_k,
+            keep_top_k=keep_top_k)
+    else:
+        return _multiclass_nms(
+            boxes,
+            scores,
+            max_output_boxes_per_class=max_output_boxes_per_class,
+            iou_threshold=iou_threshold,
+            score_threshold=score_threshold,
+            pre_top_k=pre_top_k,
+            keep_top_k=keep_top_k)
+
+
 @FUNCTION_REWRITER.register_rewriter(
     func_name='mmdeploy.codebase.mmdet.core.post_processing._multiclass_nms',
     backend='tensorrt')