[Fix] fix unittest and suppress warning

csrc/mmdeploy/backend_ops/tensorrt/roi_align/trt_roi_align.cpp

@@ -203,9 +203,9 @@ nvinfer1::IPluginV2 *TRTRoIAlignCreator::createPlugin(
       int data_size = fc->fields[i].length;
       const char *data_start = static_cast<const char *>(fc->fields[i].data);
       std::string poolModeStr(data_start, data_size);
-      if (poolModeStr == "avg") {
+      if (strcmp(poolModeStr.c_str(), "avg") == 0) {
         poolMode = 1;
-      } else if (poolModeStr == "max") {
+      } else if (strcmp(poolModeStr.c_str(), "max") == 0) {
         poolMode = 0;
       } else {
         std::cout << "Unknown pool mode \"" << poolModeStr << "\"." << std::endl;

mmdeploy/backend/tensorrt/utils.py

@@ -13,15 +13,17 @@
 from .init_plugins import load_tensorrt_plugin
 
 
-def save(engine: trt.ICudaEngine, path: str) -> None:
+def save(engine: Any, path: str) -> None:
     """Serialize TensorRT engine to disk.
 
     Args:
-        engine (tensorrt.ICudaEngine): TensorRT engine to be serialized.
+        engine (Any): TensorRT engine to be serialized.
         path (str): The absolute disk path to write the engine.
     """
     with open(path, mode='wb') as f:
-        f.write(bytearray(engine.serialize()))
+        if isinstance(engine, trt.ICudaEngine):
+            engine = engine.serialize()
+        f.write(bytearray(engine))
 
 
 def load(path: str, allocator: Optional[Any] = None) -> trt.ICudaEngine:
@@ -226,7 +228,10 @@ def from_onnx(onnx_model: Union[str, onnx.ModelProto],
             builder.int8_calibrator = config.int8_calibrator
 
     # create engine
-    engine = builder.build_engine(network, config)
+    if hasattr(builder, 'build_serialized_network'):
+        engine = builder.build_serialized_network(network, config)
+    else:
+        engine = builder.build_engine(network, config)
 
     assert engine is not None, 'Failed to create TensorRT engine'
 

mmdeploy/codebase/mmdet/deploy/object_detection_model.py

@@ -600,7 +600,7 @@ class labels of shape [N, num_det].
         scores = out[:, :, 1:2]
         boxes = out[:, :, 2:6] * scales
         dets = torch.cat([boxes, scores], dim=2)
-        return dets, torch.tensor(labels, dtype=torch.int32)
+        return dets, labels.to(torch.int32)
 
 
 @__BACKEND_MODEL.register_module('sdk')

mmdeploy/codebase/mmdet/models/backbones.py

@@ -46,7 +46,7 @@ def focus__forward__ncnn(self, x):
 
     x = x.reshape(batch_size, c * h, 1, w)
     _b, _c, _h, _w = x.shape
-    g = _c // 2
+    g = torch.div(_c, 2, rounding_mode='floor')
     # fuse to ncnn's shufflechannel
     x = x.view(_b, g, 2, _h, _w)
     x = torch.transpose(x, 1, 2).contiguous()
@@ -55,13 +55,14 @@ def focus__forward__ncnn(self, x):
     x = x.reshape(_b, c * h * w, 1, 1)
 
     _b, _c, _h, _w = x.shape
-    g = _c // 2
+    g = torch.div(_c, 2, rounding_mode='floor')
     # fuse to ncnn's shufflechannel
     x = x.view(_b, g, 2, _h, _w)
     x = torch.transpose(x, 1, 2).contiguous()
     x = x.view(_b, -1, _h, _w)
 
-    x = x.reshape(_b, c * 4, h // 2, w // 2)
+    x = x.reshape(_b, c * 4, torch.div(h, 2, rounding_mode='floor'),
+                  torch.div(w, 2, rounding_mode='floor'))
 
     return self.conv(x)
 
@@ -198,8 +199,12 @@ def shift_window_msa__forward__default(self, query, hw_shape):
         [query,
          query.new_zeros(B, C, self.window_size, query.shape[-1])],
         dim=-2)
-    slice_h = (H + self.window_size - 1) // self.window_size * self.window_size
-    slice_w = (W + self.window_size - 1) // self.window_size * self.window_size
+    slice_h = torch.div(
+        (H + self.window_size - 1), self.window_size,
+        rounding_mode='floor') * self.window_size
+    slice_w = torch.div(
+        (W + self.window_size - 1), self.window_size,
+        rounding_mode='floor') * self.window_size
     query = query[:, :, :slice_h, :slice_w]
     query = query.permute(0, 2, 3, 1).contiguous()
     H_pad, W_pad = query.shape[1], query.shape[2]

mmdeploy/core/rewriters/rewriter_utils.py

@@ -328,6 +328,29 @@ def decorator(object):
 
         return decorator
 
+    def remove_record(self, object: Any, filter_cb: Optional[Callable] = None):
+        """Remove record.
+
+        Args:
+            object (Any): The object to remove.
+            filter_cb (Callable): Check if the object need to be remove.
+                Defaults to None.
+        """
+        key_to_pop = []
+        for key, records in self._rewrite_records.items():
+            for rec in records:
+                if rec['_object'] == object:
+                    if filter_cb is not None:
+                        if filter_cb(rec):
+                            continue
+                    key_to_pop.append((key, rec))
+
+        for key, rec in key_to_pop:
+            records = self._rewrite_records[key]
+            records.remove(rec)
+            if len(records) == 0:
+                self._rewrite_records.pop(key)
+
 
 class ContextCaller:
     """A callable object used in RewriteContext.

mmdeploy/mmcv/ops/nms.py

@@ -511,7 +511,7 @@ def multiclass_nms(boxes: Tensor,
 
 
 @FUNCTION_REWRITER.register_rewriter(
-    func_name='mmdeploy.mmcv.ops.nms.bbox_nms._multiclass_nms',
+    func_name='mmdeploy.mmcv.ops.nms._multiclass_nms',
     backend=Backend.COREML.value)
 def multiclass_nms__coreml(boxes: Tensor,
                            scores: Tensor,
@@ -574,8 +574,7 @@ def _xywh2xyxy(boxes):
 
 
 @FUNCTION_REWRITER.register_rewriter(
-    func_name='mmdeploy.mmcv.ops.nms.bbox_nms._multiclass_nms',
-    ir=IR.TORCHSCRIPT)
+    func_name='mmdeploy.mmcv.ops.nms._multiclass_nms', ir=IR.TORCHSCRIPT)
 def multiclass_nms__torchscript(boxes: Tensor,
                                 scores: Tensor,
                                 max_output_boxes_per_class: int = 1000,
@@ -676,8 +675,7 @@ def symbolic(g, bboxes, scores, score_thr, iou_thr, max_size_p_class,
 
 
 @FUNCTION_REWRITER.register_rewriter(
-    func_name='mmdeploy.mmcv.ops.nms.bbox_nms._multiclass_nms',
-    backend='ascend')
+    func_name='mmdeploy.mmcv.ops.nms._multiclass_nms', backend='ascend')
 def multiclass_nms__ascend(boxes: Tensor,
                            scores: Tensor,
                            max_output_boxes_per_class: int = 1000,

mmdeploy/utils/test.py

@@ -14,6 +14,11 @@
 from mmengine.model import BaseModel
 from torch import nn
 
+try:
+    from torch.testing import assert_close as torch_assert_close
+except Exception:
+    from torch.testing import assert_allclose as torch_assert_close
+
 import mmdeploy.codebase  # noqa: F401,F403
 from mmdeploy.core import RewriterContext, patch_model
 from mmdeploy.utils import (IR, Backend, get_backend, get_dynamic_axes,
@@ -289,8 +294,7 @@ def assert_allclose(expected: List[Union[torch.Tensor, np.ndarray]],
         if isinstance(actual[i], (list, np.ndarray)):
             actual[i] = torch.tensor(actual[i])
         try:
-            torch.testing.assert_allclose(
-                actual[i], expected[i], rtol=1e-03, atol=1e-05)
+            torch_assert_close(actual[i], expected[i], rtol=1e-03, atol=1e-05)
         except AssertionError as error:
             if tolerate_small_mismatch:
                 assert '(0.00%)' in str(error), str(error)

tests/test_codebase/test_mmdet/test_mmdet_models.py

@@ -9,6 +9,12 @@
 import numpy as np
 import pytest
 import torch
+
+try:
+    from torch.testing import assert_close as torch_assert_close
+except Exception:
+    from torch.testing import assert_allclose as torch_assert_close
+
 from mmengine import Config
 from mmengine.config import ConfigDict
 
@@ -237,7 +243,7 @@ def single_level_grid_priors(input):
     # test forward
     with RewriterContext({}, backend_type):
         wrap_output = wrapped_func(x)
-        torch.testing.assert_allclose(output, wrap_output)
+        torch_assert_close(output, wrap_output)
 
     onnx_prefix = tempfile.NamedTemporaryFile().name
 
@@ -341,23 +347,6 @@ def get_ssd_head_model():
     return model
 
 
-def get_fcos_head_model():
-    """FCOS Head Config."""
-    test_cfg = Config(
-        dict(
-            deploy_nms_pre=0,
-            min_bbox_size=0,
-            score_thr=0.05,
-            nms=dict(type='nms', iou_threshold=0.5),
-            max_per_img=100))
-
-    from mmdet.models.dense_heads import FCOSHead
-    model = FCOSHead(num_classes=4, in_channels=1, test_cfg=test_cfg)
-
-    model.requires_grad_(False)
-    return model
-
-
 def get_focus_backbone_model():
     """Backbone Focus Config."""
     from mmdet.models.backbones.csp_darknet import Focus
@@ -412,10 +401,8 @@ def get_reppoints_head_model():
 
 def get_detrhead_model():
     """DETR head Config."""
-    from mmdet.models import build_head
-    from mmdet.utils import register_all_modules
-    register_all_modules()
-    model = build_head(
+    from mmdet.registry import MODELS
+    model = MODELS.build(
         dict(
             type='DETRHead',
             num_classes=4,
@@ -431,8 +418,7 @@ def get_detrhead_model():
                             dict(
                                 type='MultiheadAttention',
                                 embed_dims=4,
-                                num_heads=1,
-                                dropout=0.1)
+                                num_heads=1)
                         ],
                         ffn_cfgs=dict(
                             type='FFN',
@@ -442,8 +428,6 @@ def get_detrhead_model():
                             ffn_drop=0.,
                             act_cfg=dict(type='ReLU', inplace=True),
                         ),
-                        feedforward_channels=32,
-                        ffn_dropout=0.1,
                         operation_order=('self_attn', 'norm', 'ffn', 'norm'))),
                 decoder=dict(
                     type='DetrTransformerDecoder',
@@ -454,8 +438,7 @@ def get_detrhead_model():
                         attn_cfgs=dict(
                             type='MultiheadAttention',
                             embed_dims=4,
-                            num_heads=1,
-                            dropout=0.1),
+                            num_heads=1),
                         ffn_cfgs=dict(
                             type='FFN',
                             embed_dims=4,
@@ -465,7 +448,6 @@ def get_detrhead_model():
                             act_cfg=dict(type='ReLU', inplace=True),
                         ),
                         feedforward_channels=32,
-                        ffn_dropout=0.1,
                         operation_order=('self_attn', 'norm', 'cross_attn',
                                          'norm', 'ffn', 'norm')),
                 )),
@@ -536,7 +518,7 @@ def test_focus_forward(backend_type):
     for model_output, rewrite_output in zip(model_outputs[0], rewrite_outputs):
         model_output = model_output.squeeze()
         rewrite_output = rewrite_output.squeeze()
-        torch.testing.assert_allclose(
+        torch_assert_close(
             model_output, rewrite_output, rtol=1e-03, atol=1e-05)
 
 
@@ -578,77 +560,6 @@ def test_l2norm_forward(backend_type):
                 model_output[0], rewrite_output, rtol=1e-03, atol=1e-05)
 
 
-def test_predict_by_feat_of_fcos_head_ncnn():
-    backend_type = Backend.NCNN
-    check_backend(backend_type)
-    fcos_head = get_fcos_head_model()
-    fcos_head.cpu().eval()
-    s = 128
-    batch_img_metas = [{
-        'scale_factor': np.ones(4),
-        'pad_shape': (s, s, 3),
-        'img_shape': (s, s, 3)
-    }]
-
-    output_names = ['detection_output']
-    deploy_cfg = Config(
-        dict(
-            backend_config=dict(type=backend_type.value),
-            onnx_config=dict(output_names=output_names, input_shape=None),
-            codebase_config=dict(
-                type='mmdet',
-                task='ObjectDetection',
-                model_type='ncnn_end2end',
-                post_processing=dict(
-                    score_threshold=0.05,
-                    iou_threshold=0.5,
-                    max_output_boxes_per_class=200,
-                    pre_top_k=5000,
-                    keep_top_k=100,
-                    background_label_id=-1,
-                ))))
-
-    # the cls_score's size: (1, 36, 32, 32), (1, 36, 16, 16),
-    # (1, 36, 8, 8), (1, 36, 4, 4), (1, 36, 2, 2).
-    # the bboxes's size: (1, 36, 32, 32), (1, 36, 16, 16),
-    # (1, 36, 8, 8), (1, 36, 4, 4), (1, 36, 2, 2)
-    seed_everything(1234)
-    cls_score = [
-        torch.rand(1, fcos_head.num_classes, pow(2, i), pow(2, i))
-        for i in range(5, 0, -1)
-    ]
-    seed_everything(5678)
-    bboxes = [torch.rand(1, 4, pow(2, i), pow(2, i)) for i in range(5, 0, -1)]
-
-    seed_everything(9101)
-    centernesses = [
-        torch.rand(1, 1, pow(2, i), pow(2, i)) for i in range(5, 0, -1)
-    ]
-
-    # to get outputs of onnx model after rewrite
-    batch_img_metas[0]['img_shape'] = torch.Tensor([s, s])
-    wrapped_model = WrapModel(
-        fcos_head,
-        'predict_by_feat',
-        batch_img_metas=batch_img_metas,
-        with_nms=True)
-    rewrite_inputs = {
-        'cls_scores': cls_score,
-        'bbox_preds': bboxes,
-        'centernesses': centernesses
-    }
-    rewrite_outputs, is_backend_output = get_rewrite_outputs(
-        wrapped_model=wrapped_model,
-        model_inputs=rewrite_inputs,
-        deploy_cfg=deploy_cfg)
-
-    # output should be of shape [1, N, 6]
-    if is_backend_output:
-        assert rewrite_outputs[0].shape[-1] == 6
-    else:
-        assert rewrite_outputs.shape[-1] == 6
-
-
 @pytest.mark.parametrize('backend_type', [Backend.ONNXRUNTIME, Backend.NCNN])
 def test_predict_by_feat_of_rpn_head(backend_type: Backend):
     check_backend(backend_type)

tests/test_codebase/test_mmdet3d/test_voxel_detection_model.py

@@ -57,6 +57,10 @@ def setup_class(cls):
             deploy_cfg=deploy_cfg,
             model_cfg=model_cfg)
 
+    @classmethod
+    def teardown_class(cls):
+        cls.wrapper.recover()
+
     @pytest.mark.skipif(
         reason='Only support GPU test',
         condition=not torch.cuda.is_available())