Export, detect and validation with TensorRT engine file (#5699)

* Export and detect with TensorRT engine file

* Resolve `isort`

* Make validation works with TensorRT engine

* feat: update export docstring

* feat: change suffix from *.trt to *.engine

* feat: get rid of pycuda

* feat: make compatiable with val.py

* feat: support detect with fp16 engine

* Add Lite to Edge TPU string

* Remove *.trt comment

* Revert to standard success logger.info string

* Fix Deprecation Warning

```
export.py:310: DeprecationWarning: Use build_serialized_network instead.
  with builder.build_engine(network, config) as engine, open(f, 'wb') as t:
```

* Revert deprecation warning fix

@imyhxy it seems we can't apply the deprecation warning fix because then export fails, so I'm reverting my previous change here.

* Update export.py

* Update export.py

* Update common.py

* export onnx to file before building TensorRT engine file

* feat: triger ONNX export failed early

* feat: load ONNX model from file

Co-authored-by: Glenn Jocher <glenn.jocher@ultralytics.com>

detect.py

     # Load model
     device = select_device(device)
     model = DetectMultiBackend(weights, device=device, dnn=dnn)
-    stride, names, pt, jit, onnx = model.stride, model.names, model.pt, model.jit, model.onnx
+    stride, names, pt, jit, onnx, engine = model.stride, model.names, model.pt, model.jit, model.onnx, model.engine
     imgsz = check_img_size(imgsz, s=stride)  # check image size
 
     # Half
-    half &= pt and device.type != 'cpu'  # half precision only supported by PyTorch on CUDA
+    half &= (pt or engine) and device.type != 'cpu'  # half precision only supported by PyTorch on CUDA
     if pt:
         model.model.half() if half else model.model.float()
 

export.py

 TensorFlow GraphDef     | yolov5s.pb                | 'pb'
 TensorFlow Lite         | yolov5s.tflite            | 'tflite'
 TensorFlow.js           | yolov5s_web_model/        | 'tfjs'
+TensorRT                | yolov5s.engine            | 'engine'
 
 Usage:
     $ python path/to/export.py --weights yolov5s.pt --include torchscript onnx coreml saved_model pb tflite tfjs
                                          yolov5s_saved_model
                                          yolov5s.pb
                                          yolov5s.tflite
+                                         yolov5s.engine
 
 TensorFlow.js:
     $ cd .. && git clone https://github.com/zldrobit/tfjs-yolov5-example.git && cd tfjs-yolov5-example
         LOGGER.info(f'\n{prefix} export failure: {e}')
 
 
+def export_engine(model, im, file, train, half, simplify, workspace=4, verbose=False, prefix=colorstr('TensorRT:')):
+    try:
+        check_requirements(('tensorrt',))
+        import tensorrt as trt
+
+        opset = (12, 13)[trt.__version__[0] == '8']  # test on TensorRT 7.x and 8.x
+        export_onnx(model, im, file, opset, train, False, simplify)
+        onnx = file.with_suffix('.onnx')
+        assert onnx.exists(), f'failed to export ONNX file: {onnx}'
+
+        LOGGER.info(f'\n{prefix} starting export with TensorRT {trt.__version__}...')
+        f = str(file).replace('.pt', '.engine')  # TensorRT engine file
+        logger = trt.Logger(trt.Logger.INFO)
+        if verbose:
+            logger.min_severity = trt.Logger.Severity.VERBOSE
+
+        builder = trt.Builder(logger)
+        config = builder.create_builder_config()
+        config.max_workspace_size = workspace * 1 << 30
+
+        flag = (1 << int(trt.NetworkDefinitionCreationFlag.EXPLICIT_BATCH))
+        network = builder.create_network(flag)
+        parser = trt.OnnxParser(network, logger)
+        if not parser.parse_from_file(str(onnx)):
+            raise RuntimeError(f'failed to load ONNX file: {onnx}')
+
+        inputs = [network.get_input(i) for i in range(network.num_inputs)]
+        outputs = [network.get_output(i) for i in range(network.num_outputs)]
+        LOGGER.info(f'{prefix} Network Description:')
+        for inp in inputs:
+            LOGGER.info(f'{prefix}\tinput "{inp.name}" with shape {inp.shape} and dtype {inp.dtype}')
+        for out in outputs:
+            LOGGER.info(f'{prefix}\toutput "{out.name}" with shape {out.shape} and dtype {out.dtype}')
+
+        half &= builder.platform_has_fast_fp16
+        LOGGER.info(f'{prefix} building FP{16 if half else 32} engine in {f}')
+        if half:
+            config.set_flag(trt.BuilderFlag.FP16)
+        with builder.build_engine(network, config) as engine, open(f, 'wb') as t:
+            t.write(engine.serialize())
+        LOGGER.info(f'{prefix} export success, saved as {f} ({file_size(f):.1f} MB)')
+
+    except Exception as e:
+        LOGGER.info(f'\n{prefix} export failure: {e}')
+
 @torch.no_grad()
 def run(data=ROOT / 'data/coco128.yaml',  # 'dataset.yaml path'
         weights=ROOT / 'yolov5s.pt',  # weights path
         dynamic=False,  # ONNX/TF: dynamic axes
         simplify=False,  # ONNX: simplify model
         opset=12,  # ONNX: opset version
+        verbose=False,  # TensorRT: verbose log
+        workspace=4,  # TensorRT: workspace size (GB)
         topk_per_class=100,  # TF.js NMS: topk per class to keep
         topk_all=100,  # TF.js NMS: topk for all classes to keep
         iou_thres=0.45,  # TF.js NMS: IoU threshold
         export_torchscript(model, im, file, optimize)
     if 'onnx' in include:
         export_onnx(model, im, file, opset, train, dynamic, simplify)
+    if 'engine' in include:
+        export_engine(model, im, file, train, half, simplify, workspace, verbose)
     if 'coreml' in include:
         export_coreml(model, im, file)
 
     parser.add_argument('--dynamic', action='store_true', help='ONNX/TF: dynamic axes')
     parser.add_argument('--simplify', action='store_true', help='ONNX: simplify model')
     parser.add_argument('--opset', type=int, default=13, help='ONNX: opset version')
+    parser.add_argument('--verbose', action='store_true', help='TensorRT: verbose log')
+    parser.add_argument('--workspace', type=int, default=4, help='TensorRT: workspace size (GB)')
     parser.add_argument('--topk-per-class', type=int, default=100, help='TF.js NMS: topk per class to keep')
     parser.add_argument('--topk-all', type=int, default=100, help='TF.js NMS: topk for all classes to keep')
     parser.add_argument('--iou-thres', type=float, default=0.45, help='TF.js NMS: IoU threshold')
     parser.add_argument('--conf-thres', type=float, default=0.25, help='TF.js NMS: confidence threshold')
     parser.add_argument('--include', nargs='+',
                         default=['torchscript', 'onnx'],
-                        help='available formats are (torchscript, onnx, coreml, saved_model, pb, tflite, tfjs)')
+                        help='available formats are (torchscript, onnx, engine, coreml, saved_model, pb, tflite, tfjs)')
     opt = parser.parse_args()
     print_args(FILE.stem, opt)
     return opt

models/common.py

 import math
 import platform
 import warnings
+from collections import namedtuple
 from copy import copy
 from pathlib import Path
 
         #   TensorFlow Lite:        *.tflite
         #   ONNX Runtime:           *.onnx
         #   OpenCV DNN:             *.onnx with dnn=True
+        #   TensorRT:               *.engine
         super().__init__()
         w = str(weights[0] if isinstance(weights, list) else weights)
-        suffix, suffixes = Path(w).suffix.lower(), ['.pt', '.onnx', '.tflite', '.pb', '', '.mlmodel']
+        suffix, suffixes = Path(w).suffix.lower(), ['.pt', '.onnx', '.engine', '.tflite', '.pb', '', '.mlmodel']
         check_suffix(w, suffixes)  # check weights have acceptable suffix
-        pt, onnx, tflite, pb, saved_model, coreml = (suffix == x for x in suffixes)  # backend booleans
+        pt, onnx, engine, tflite, pb, saved_model, coreml = (suffix == x for x in suffixes)  # backend booleans
         jit = pt and 'torchscript' in w.lower()
         stride, names = 64, [f'class{i}' for i in range(1000)]  # assign defaults
 
             check_requirements(('onnx', 'onnxruntime-gpu' if torch.has_cuda else 'onnxruntime'))
             import onnxruntime
             session = onnxruntime.InferenceSession(w, None)
+        elif engine:  # TensorRT
+            LOGGER.info(f'Loading {w} for TensorRT inference...')
+            import tensorrt as trt  # https://developer.nvidia.com/nvidia-tensorrt-download
+            Binding = namedtuple('Binding', ('name', 'dtype', 'shape', 'data', 'ptr'))
+            logger = trt.Logger(trt.Logger.INFO)
+            with open(w, 'rb') as f, trt.Runtime(logger) as runtime:
+                model = runtime.deserialize_cuda_engine(f.read())
+            bindings = dict()
+            for index in range(model.num_bindings):
+                name = model.get_binding_name(index)
+                dtype = trt.nptype(model.get_binding_dtype(index))
+                shape = tuple(model.get_binding_shape(index))
+                data = torch.from_numpy(np.empty(shape, dtype=np.dtype(dtype))).to(device)
+                bindings[name] = Binding(name, dtype, shape, data, int(data.data_ptr()))
+            binding_addrs = {n: d.ptr for n, d in bindings.items()}
+            context = model.create_execution_context()
+            batch_size = bindings['images'].shape[0]
         else:  # TensorFlow model (TFLite, pb, saved_model)
             import tensorflow as tf
             if pb:  # https://www.tensorflow.org/guide/migrate#a_graphpb_or_graphpbtxt
                 model = tf.keras.models.load_model(w)
             elif tflite:  # https://www.tensorflow.org/lite/guide/python#install_tensorflow_lite_for_python
                 if 'edgetpu' in w.lower():
-                    LOGGER.info(f'Loading {w} for TensorFlow Edge TPU inference...')
+                    LOGGER.info(f'Loading {w} for TensorFlow Lite Edge TPU inference...')
                     import tflite_runtime.interpreter as tfli
                     delegate = {'Linux': 'libedgetpu.so.1',  # install https://coral.ai/software/#edgetpu-runtime
                                 'Darwin': 'libedgetpu.1.dylib',
                 y = self.net.forward()
             else:  # ONNX Runtime
                 y = self.session.run([self.session.get_outputs()[0].name], {self.session.get_inputs()[0].name: im})[0]
+        elif self.engine:  # TensorRT
+            assert im.shape == self.bindings['images'].shape, (im.shape, self.bindings['images'].shape)
+            self.binding_addrs['images'] = int(im.data_ptr())
+            self.context.execute_v2(list(self.binding_addrs.values()))
+            y = self.bindings['output'].data
         else:  # TensorFlow model (TFLite, pb, saved_model)
             im = im.permute(0, 2, 3, 1).cpu().numpy()  # torch BCHW to numpy BHWC shape(1,320,192,3)
             if self.pb:
             y[..., 1] *= h  # y
             y[..., 2] *= w  # w
             y[..., 3] *= h  # h
-        y = torch.tensor(y)
+        y = torch.tensor(y) if isinstance(y, np.ndarray) else y
         return (y, []) if val else y
 
 

val.py

     # Initialize/load model and set device
     training = model is not None
     if training:  # called by train.py
-        device, pt = next(model.parameters()).device, True  # get model device, PyTorch model
+        device, pt, engine = next(model.parameters()).device, True, False  # get model device, PyTorch model
 
         half &= device.type != 'cpu'  # half precision only supported on CUDA
         model.half() if half else model.float()
 
         # Load model
         model = DetectMultiBackend(weights, device=device, dnn=dnn)
-        stride, pt = model.stride, model.pt
+        stride, pt, engine = model.stride, model.pt, model.engine
         imgsz = check_img_size(imgsz, s=stride)  # check image size
-        half &= pt and device.type != 'cpu'  # half precision only supported by PyTorch on CUDA
+        half &= (pt or engine) and device.type != 'cpu'  # half precision only supported by PyTorch on CUDA
         if pt:
             model.model.half() if half else model.model.float()
+        elif engine:
+            batch_size = model.batch_size
         else:
             half = False
             batch_size = 1  # export.py models default to batch-size 1
     pbar = tqdm(dataloader, desc=s, ncols=NCOLS, bar_format='{l_bar}{bar:10}{r_bar}{bar:-10b}')  # progress bar
     for batch_i, (im, targets, paths, shapes) in enumerate(pbar):
         t1 = time_sync()
-        if pt:
+        if pt or engine:
             im = im.to(device, non_blocking=True)
             targets = targets.to(device)
         im = im.half() if half else im.float()  # uint8 to fp16/32