File size: 41,190 Bytes
30c8b41
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
# YOLOv5 🚀 by Ultralytics, AGPL-3.0 license
"""
Export a YOLOv5 PyTorch model to other formats. TensorFlow exports authored by https://github.com/zldrobit

Format                      | `export.py --include`         | Model
---                         | ---                           | ---
PyTorch                     | -                             | yolov5s.pt
TorchScript                 | `torchscript`                 | yolov5s.torchscript
ONNX                        | `onnx`                        | yolov5s.onnx
OpenVINO                    | `openvino`                    | yolov5s_openvino_model/
TensorRT                    | `engine`                      | yolov5s.engine
CoreML                      | `coreml`                      | yolov5s.mlmodel
TensorFlow SavedModel       | `saved_model`                 | yolov5s_saved_model/
TensorFlow GraphDef         | `pb`                          | yolov5s.pb
TensorFlow Lite             | `tflite`                      | yolov5s.tflite
TensorFlow Edge TPU         | `edgetpu`                     | yolov5s_edgetpu.tflite
TensorFlow.js               | `tfjs`                        | yolov5s_web_model/
PaddlePaddle                | `paddle`                      | yolov5s_paddle_model/

Requirements:
    $ pip install -r requirements.txt coremltools onnx onnx-simplifier onnxruntime openvino-dev tensorflow-cpu  # CPU
    $ pip install -r requirements.txt coremltools onnx onnx-simplifier onnxruntime-gpu openvino-dev tensorflow  # GPU

Usage:
    $ python export.py --weights yolov5s.pt --include torchscript onnx openvino engine coreml tflite ...

Inference:
    $ python detect.py --weights yolov5s.pt                 # PyTorch
                                 yolov5s.torchscript        # TorchScript
                                 yolov5s.onnx               # ONNX Runtime or OpenCV DNN with --dnn
                                 yolov5s_openvino_model     # OpenVINO
                                 yolov5s.engine             # TensorRT
                                 yolov5s.mlmodel            # CoreML (macOS-only)
                                 yolov5s_saved_model        # TensorFlow SavedModel
                                 yolov5s.pb                 # TensorFlow GraphDef
                                 yolov5s.tflite             # TensorFlow Lite
                                 yolov5s_edgetpu.tflite     # TensorFlow Edge TPU
                                 yolov5s_paddle_model       # PaddlePaddle

TensorFlow.js:
    $ cd .. && git clone https://github.com/zldrobit/tfjs-yolov5-example.git && cd tfjs-yolov5-example
    $ npm install
    $ ln -s ../../yolov5/yolov5s_web_model public/yolov5s_web_model
    $ npm start
"""

import argparse
import contextlib
import json
import os
import platform
import re
import subprocess
import sys
import time
import warnings
from pathlib import Path

import pandas as pd
import torch
from torch.utils.mobile_optimizer import optimize_for_mobile

FILE = Path(__file__).resolve()
ROOT = FILE.parents[0]  # YOLOv5 root directory
if str(ROOT) not in sys.path:
    sys.path.append(str(ROOT))  # add ROOT to PATH
if platform.system() != 'Windows':
    ROOT = Path(os.path.relpath(ROOT, Path.cwd()))  # relative

from models.experimental import attempt_load
from models.yolo import ClassificationModel, Detect, DetectionModel, SegmentationModel
from utils.dataloaders import LoadImages
from utils.general import (LOGGER, Profile, check_dataset, check_img_size, check_requirements, check_version,
                           check_yaml, colorstr, file_size, get_default_args, print_args, url2file, yaml_save)
from utils.torch_utils import select_device, smart_inference_mode

MACOS = platform.system() == 'Darwin'  # macOS environment


class iOSModel(torch.nn.Module):

    def __init__(self, model, im):
        super().__init__()
        b, c, h, w = im.shape  # batch, channel, height, width
        self.model = model
        self.nc = model.nc  # number of classes
        if w == h:
            self.normalize = 1. / w
        else:
            self.normalize = torch.tensor([1. / w, 1. / h, 1. / w, 1. / h])  # broadcast (slower, smaller)
            # np = model(im)[0].shape[1]  # number of points
            # self.normalize = torch.tensor([1. / w, 1. / h, 1. / w, 1. / h]).expand(np, 4)  # explicit (faster, larger)

    def forward(self, x):
        xywh, conf, cls = self.model(x)[0].squeeze().split((4, 1, self.nc), 1)
        return cls * conf, xywh * self.normalize  # confidence (3780, 80), coordinates (3780, 4)


def export_formats():
    # YOLOv5 export formats
    x = [
        ['PyTorch', '-', '.pt', True, True],
        ['TorchScript', 'torchscript', '.torchscript', True, True],
        ['ONNX', 'onnx', '.onnx', True, True],
        ['OpenVINO', 'openvino', '_openvino_model', True, False],
        ['TensorRT', 'engine', '.engine', False, True],
        ['CoreML', 'coreml', '.mlmodel', True, False],
        ['TensorFlow SavedModel', 'saved_model', '_saved_model', True, True],
        ['TensorFlow GraphDef', 'pb', '.pb', True, True],
        ['TensorFlow Lite', 'tflite', '.tflite', True, False],
        ['TensorFlow Edge TPU', 'edgetpu', '_edgetpu.tflite', False, False],
        ['TensorFlow.js', 'tfjs', '_web_model', False, False],
        ['PaddlePaddle', 'paddle', '_paddle_model', True, True], ]
    return pd.DataFrame(x, columns=['Format', 'Argument', 'Suffix', 'CPU', 'GPU'])


def try_export(inner_func):
    # YOLOv5 export decorator, i..e @try_export
    inner_args = get_default_args(inner_func)

    def outer_func(*args, **kwargs):
        prefix = inner_args['prefix']
        try:
            with Profile() as dt:
                f, model = inner_func(*args, **kwargs)
            LOGGER.info(f'{prefix} export success ✅ {dt.t:.1f}s, saved as {f} ({file_size(f):.1f} MB)')
            return f, model
        except Exception as e:
            LOGGER.info(f'{prefix} export failure ❌ {dt.t:.1f}s: {e}')
            return None, None

    return outer_func


@try_export
def export_torchscript(model, im, file, optimize, prefix=colorstr('TorchScript:')):
    # YOLOv5 TorchScript model export
    LOGGER.info(f'\n{prefix} starting export with torch {torch.__version__}...')
    f = file.with_suffix('.torchscript')

    ts = torch.jit.trace(model, im, strict=False)
    d = {'shape': im.shape, 'stride': int(max(model.stride)), 'names': model.names}
    extra_files = {'config.txt': json.dumps(d)}  # torch._C.ExtraFilesMap()
    if optimize:  # https://pytorch.org/tutorials/recipes/mobile_interpreter.html
        optimize_for_mobile(ts)._save_for_lite_interpreter(str(f), _extra_files=extra_files)
    else:
        ts.save(str(f), _extra_files=extra_files)
    return f, None


@try_export
def export_onnx(model, im, file, opset, dynamic, simplify, prefix=colorstr('ONNX:')):
    # YOLOv5 ONNX export
    check_requirements('onnx>=1.12.0')
    import onnx

    LOGGER.info(f'\n{prefix} starting export with onnx {onnx.__version__}...')
    f = file.with_suffix('.onnx')

    output_names = ['output0', 'output1'] if isinstance(model, SegmentationModel) else ['output0']
    if dynamic:
        dynamic = {'images': {0: 'batch', 2: 'height', 3: 'width'}}  # shape(1,3,640,640)
        if isinstance(model, SegmentationModel):
            dynamic['output0'] = {0: 'batch', 1: 'anchors'}  # shape(1,25200,85)
            dynamic['output1'] = {0: 'batch', 2: 'mask_height', 3: 'mask_width'}  # shape(1,32,160,160)
        elif isinstance(model, DetectionModel):
            dynamic['output0'] = {0: 'batch', 1: 'anchors'}  # shape(1,25200,85)

    torch.onnx.export(
        model.cpu() if dynamic else model,  # --dynamic only compatible with cpu
        im.cpu() if dynamic else im,
        f,
        verbose=False,
        opset_version=opset,
        do_constant_folding=True,  # WARNING: DNN inference with torch>=1.12 may require do_constant_folding=False
        input_names=['images'],
        output_names=output_names,
        dynamic_axes=dynamic or None)

    # Checks
    model_onnx = onnx.load(f)  # load onnx model
    onnx.checker.check_model(model_onnx)  # check onnx model

    # Metadata
    d = {'stride': int(max(model.stride)), 'names': model.names}
    for k, v in d.items():
        meta = model_onnx.metadata_props.add()
        meta.key, meta.value = k, str(v)
    onnx.save(model_onnx, f)

    # Simplify
    if simplify:
        try:
            cuda = torch.cuda.is_available()
            check_requirements(('onnxruntime-gpu' if cuda else 'onnxruntime', 'onnx-simplifier>=0.4.1'))
            import onnxsim

            LOGGER.info(f'{prefix} simplifying with onnx-simplifier {onnxsim.__version__}...')
            model_onnx, check = onnxsim.simplify(model_onnx)
            assert check, 'assert check failed'
            onnx.save(model_onnx, f)
        except Exception as e:
            LOGGER.info(f'{prefix} simplifier failure: {e}')
    return f, model_onnx


@try_export
def export_openvino(file, metadata, half, int8, data, prefix=colorstr('OpenVINO:')):
    # YOLOv5 OpenVINO export
    check_requirements('openvino-dev>=2022.3')  # requires openvino-dev: https://pypi.org/project/openvino-dev/
    import openvino.runtime as ov  # noqa
    from openvino.tools import mo  # noqa

    LOGGER.info(f'\n{prefix} starting export with openvino {ov.__version__}...')
    f = str(file).replace(file.suffix, f'_openvino_model{os.sep}')
    f_onnx = file.with_suffix('.onnx')
    f_ov = str(Path(f) / file.with_suffix('.xml').name)
    if int8:
        check_requirements('nncf')
        import nncf
        import numpy as np
        from openvino.runtime import Core

        from utils.dataloaders import create_dataloader
        core = Core()
        onnx_model = core.read_model(f_onnx)  # export

        def prepare_input_tensor(image: np.ndarray):
            input_tensor = image.astype(np.float32)  # uint8 to fp16/32
            input_tensor /= 255.0  # 0 - 255 to 0.0 - 1.0

            if input_tensor.ndim == 3:
                input_tensor = np.expand_dims(input_tensor, 0)
            return input_tensor

        def gen_dataloader(yaml_path, task='train', imgsz=640, workers=4):
            data_yaml = check_yaml(yaml_path)
            data = check_dataset(data_yaml)
            dataloader = create_dataloader(data[task],
                                           imgsz=imgsz,
                                           batch_size=1,
                                           stride=32,
                                           pad=0.5,
                                           single_cls=False,
                                           rect=False,
                                           workers=workers)[0]
            return dataloader

        # noqa: F811

        def transform_fn(data_item):
            """
            Quantization transform function. Extracts and preprocess input data from dataloader item for quantization.
            Parameters:
               data_item: Tuple with data item produced by DataLoader during iteration
            Returns:
                input_tensor: Input data for quantization
            """
            img = data_item[0].numpy()
            input_tensor = prepare_input_tensor(img)
            return input_tensor

        ds = gen_dataloader(data)
        quantization_dataset = nncf.Dataset(ds, transform_fn)
        ov_model = nncf.quantize(onnx_model, quantization_dataset, preset=nncf.QuantizationPreset.MIXED)
    else:
        ov_model = mo.convert_model(f_onnx, model_name=file.stem, framework='onnx', compress_to_fp16=half)  # export

    ov.serialize(ov_model, f_ov)  # save
    yaml_save(Path(f) / file.with_suffix('.yaml').name, metadata)  # add metadata.yaml
    return f, None


@try_export
def export_paddle(model, im, file, metadata, prefix=colorstr('PaddlePaddle:')):
    # YOLOv5 Paddle export
    check_requirements(('paddlepaddle', 'x2paddle'))
    import x2paddle
    from x2paddle.convert import pytorch2paddle

    LOGGER.info(f'\n{prefix} starting export with X2Paddle {x2paddle.__version__}...')
    f = str(file).replace('.pt', f'_paddle_model{os.sep}')

    pytorch2paddle(module=model, save_dir=f, jit_type='trace', input_examples=[im])  # export
    yaml_save(Path(f) / file.with_suffix('.yaml').name, metadata)  # add metadata.yaml
    return f, None


@try_export
def export_coreml(model, im, file, int8, half, nms, prefix=colorstr('CoreML:')):
    # YOLOv5 CoreML export
    check_requirements('coremltools')
    import coremltools as ct

    LOGGER.info(f'\n{prefix} starting export with coremltools {ct.__version__}...')
    f = file.with_suffix('.mlmodel')

    if nms:
        model = iOSModel(model, im)
    ts = torch.jit.trace(model, im, strict=False)  # TorchScript model
    ct_model = ct.convert(ts, inputs=[ct.ImageType('image', shape=im.shape, scale=1 / 255, bias=[0, 0, 0])])
    bits, mode = (8, 'kmeans_lut') if int8 else (16, 'linear') if half else (32, None)
    if bits < 32:
        if MACOS:  # quantization only supported on macOS
            with warnings.catch_warnings():
                warnings.filterwarnings('ignore', category=DeprecationWarning)  # suppress numpy==1.20 float warning
                ct_model = ct.models.neural_network.quantization_utils.quantize_weights(ct_model, bits, mode)
        else:
            print(f'{prefix} quantization only supported on macOS, skipping...')
    ct_model.save(f)
    return f, ct_model


@try_export
def export_engine(model, im, file, half, dynamic, simplify, workspace=4, verbose=False, prefix=colorstr('TensorRT:')):
    # YOLOv5 TensorRT export https://developer.nvidia.com/tensorrt
    assert im.device.type != 'cpu', 'export running on CPU but must be on GPU, i.e. `python export.py --device 0`'
    try:
        import tensorrt as trt
    except Exception:
        if platform.system() == 'Linux':
            check_requirements('nvidia-tensorrt', cmds='-U --index-url https://pypi.ngc.nvidia.com')
        import tensorrt as trt

    if trt.__version__[0] == '7':  # TensorRT 7 handling https://github.com/ultralytics/yolov5/issues/6012
        grid = model.model[-1].anchor_grid
        model.model[-1].anchor_grid = [a[..., :1, :1, :] for a in grid]
        export_onnx(model, im, file, 12, dynamic, simplify)  # opset 12
        model.model[-1].anchor_grid = grid
    else:  # TensorRT >= 8
        check_version(trt.__version__, '8.0.0', hard=True)  # require tensorrt>=8.0.0
        export_onnx(model, im, file, 12, dynamic, simplify)  # opset 12
    onnx = file.with_suffix('.onnx')

    LOGGER.info(f'\n{prefix} starting export with TensorRT {trt.__version__}...')
    assert onnx.exists(), f'failed to export ONNX file: {onnx}'
    f = file.with_suffix('.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
    # config.set_memory_pool_limit(trt.MemoryPoolType.WORKSPACE, workspace << 30)  # fix TRT 8.4 deprecation notice

    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)]
    for inp in inputs:
        LOGGER.info(f'{prefix} input "{inp.name}" with shape{inp.shape} {inp.dtype}')
    for out in outputs:
        LOGGER.info(f'{prefix} output "{out.name}" with shape{out.shape} {out.dtype}')

    if dynamic:
        if im.shape[0] <= 1:
            LOGGER.warning(f'{prefix} WARNING ⚠️ --dynamic model requires maximum --batch-size argument')
        profile = builder.create_optimization_profile()
        for inp in inputs:
            profile.set_shape(inp.name, (1, *im.shape[1:]), (max(1, im.shape[0] // 2), *im.shape[1:]), im.shape)
        config.add_optimization_profile(profile)

    LOGGER.info(f'{prefix} building FP{16 if builder.platform_has_fast_fp16 and half else 32} engine as {f}')
    if builder.platform_has_fast_fp16 and half:
        config.set_flag(trt.BuilderFlag.FP16)
    with builder.build_engine(network, config) as engine, open(f, 'wb') as t:
        t.write(engine.serialize())
    return f, None


@try_export
def export_saved_model(model,
                       im,
                       file,
                       dynamic,
                       tf_nms=False,
                       agnostic_nms=False,
                       topk_per_class=100,
                       topk_all=100,
                       iou_thres=0.45,
                       conf_thres=0.25,
                       keras=False,
                       prefix=colorstr('TensorFlow SavedModel:')):
    # YOLOv5 TensorFlow SavedModel export
    try:
        import tensorflow as tf
    except Exception:
        check_requirements(f"tensorflow{'' if torch.cuda.is_available() else '-macos' if MACOS else '-cpu'}")
        import tensorflow as tf
    from tensorflow.python.framework.convert_to_constants import convert_variables_to_constants_v2

    from models.tf import TFModel

    LOGGER.info(f'\n{prefix} starting export with tensorflow {tf.__version__}...')
    f = str(file).replace('.pt', '_saved_model')
    batch_size, ch, *imgsz = list(im.shape)  # BCHW

    tf_model = TFModel(cfg=model.yaml, model=model, nc=model.nc, imgsz=imgsz)
    im = tf.zeros((batch_size, *imgsz, ch))  # BHWC order for TensorFlow
    _ = tf_model.predict(im, tf_nms, agnostic_nms, topk_per_class, topk_all, iou_thres, conf_thres)
    inputs = tf.keras.Input(shape=(*imgsz, ch), batch_size=None if dynamic else batch_size)
    outputs = tf_model.predict(inputs, tf_nms, agnostic_nms, topk_per_class, topk_all, iou_thres, conf_thres)
    keras_model = tf.keras.Model(inputs=inputs, outputs=outputs)
    keras_model.trainable = False
    keras_model.summary()
    if keras:
        keras_model.save(f, save_format='tf')
    else:
        spec = tf.TensorSpec(keras_model.inputs[0].shape, keras_model.inputs[0].dtype)
        m = tf.function(lambda x: keras_model(x))  # full model
        m = m.get_concrete_function(spec)
        frozen_func = convert_variables_to_constants_v2(m)
        tfm = tf.Module()
        tfm.__call__ = tf.function(lambda x: frozen_func(x)[:4] if tf_nms else frozen_func(x), [spec])
        tfm.__call__(im)
        tf.saved_model.save(tfm,
                            f,
                            options=tf.saved_model.SaveOptions(experimental_custom_gradients=False) if check_version(
                                tf.__version__, '2.6') else tf.saved_model.SaveOptions())
    return f, keras_model


@try_export
def export_pb(keras_model, file, prefix=colorstr('TensorFlow GraphDef:')):
    # YOLOv5 TensorFlow GraphDef *.pb export https://github.com/leimao/Frozen_Graph_TensorFlow
    import tensorflow as tf
    from tensorflow.python.framework.convert_to_constants import convert_variables_to_constants_v2

    LOGGER.info(f'\n{prefix} starting export with tensorflow {tf.__version__}...')
    f = file.with_suffix('.pb')

    m = tf.function(lambda x: keras_model(x))  # full model
    m = m.get_concrete_function(tf.TensorSpec(keras_model.inputs[0].shape, keras_model.inputs[0].dtype))
    frozen_func = convert_variables_to_constants_v2(m)
    frozen_func.graph.as_graph_def()
    tf.io.write_graph(graph_or_graph_def=frozen_func.graph, logdir=str(f.parent), name=f.name, as_text=False)
    return f, None


@try_export
def export_tflite(keras_model, im, file, int8, data, nms, agnostic_nms, prefix=colorstr('TensorFlow Lite:')):
    # YOLOv5 TensorFlow Lite export
    import tensorflow as tf

    LOGGER.info(f'\n{prefix} starting export with tensorflow {tf.__version__}...')
    batch_size, ch, *imgsz = list(im.shape)  # BCHW
    f = str(file).replace('.pt', '-fp16.tflite')

    converter = tf.lite.TFLiteConverter.from_keras_model(keras_model)
    converter.target_spec.supported_ops = [tf.lite.OpsSet.TFLITE_BUILTINS]
    converter.target_spec.supported_types = [tf.float16]
    converter.optimizations = [tf.lite.Optimize.DEFAULT]
    if int8:
        from models.tf import representative_dataset_gen
        dataset = LoadImages(check_dataset(check_yaml(data))['train'], img_size=imgsz, auto=False)
        converter.representative_dataset = lambda: representative_dataset_gen(dataset, ncalib=100)
        converter.target_spec.supported_ops = [tf.lite.OpsSet.TFLITE_BUILTINS_INT8]
        converter.target_spec.supported_types = []
        converter.inference_input_type = tf.uint8  # or tf.int8
        converter.inference_output_type = tf.uint8  # or tf.int8
        converter.experimental_new_quantizer = True
        f = str(file).replace('.pt', '-int8.tflite')
    if nms or agnostic_nms:
        converter.target_spec.supported_ops.append(tf.lite.OpsSet.SELECT_TF_OPS)

    tflite_model = converter.convert()
    open(f, 'wb').write(tflite_model)
    return f, None


@try_export
def export_edgetpu(file, prefix=colorstr('Edge TPU:')):
    # YOLOv5 Edge TPU export https://coral.ai/docs/edgetpu/models-intro/
    cmd = 'edgetpu_compiler --version'
    help_url = 'https://coral.ai/docs/edgetpu/compiler/'
    assert platform.system() == 'Linux', f'export only supported on Linux. See {help_url}'
    if subprocess.run(f'{cmd} > /dev/null 2>&1', shell=True).returncode != 0:
        LOGGER.info(f'\n{prefix} export requires Edge TPU compiler. Attempting install from {help_url}')
        sudo = subprocess.run('sudo --version >/dev/null', shell=True).returncode == 0  # sudo installed on system
        for c in (
                'curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add -',
                'echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | sudo tee /etc/apt/sources.list.d/coral-edgetpu.list',
                'sudo apt-get update', 'sudo apt-get install edgetpu-compiler'):
            subprocess.run(c if sudo else c.replace('sudo ', ''), shell=True, check=True)
    ver = subprocess.run(cmd, shell=True, capture_output=True, check=True).stdout.decode().split()[-1]

    LOGGER.info(f'\n{prefix} starting export with Edge TPU compiler {ver}...')
    f = str(file).replace('.pt', '-int8_edgetpu.tflite')  # Edge TPU model
    f_tfl = str(file).replace('.pt', '-int8.tflite')  # TFLite model

    subprocess.run([
        'edgetpu_compiler',
        '-s',
        '-d',
        '-k',
        '10',
        '--out_dir',
        str(file.parent),
        f_tfl, ], check=True)
    return f, None


@try_export
def export_tfjs(file, int8, prefix=colorstr('TensorFlow.js:')):
    # YOLOv5 TensorFlow.js export
    check_requirements('tensorflowjs')
    import tensorflowjs as tfjs

    LOGGER.info(f'\n{prefix} starting export with tensorflowjs {tfjs.__version__}...')
    f = str(file).replace('.pt', '_web_model')  # js dir
    f_pb = file.with_suffix('.pb')  # *.pb path
    f_json = f'{f}/model.json'  # *.json path

    args = [
        'tensorflowjs_converter',
        '--input_format=tf_frozen_model',
        '--quantize_uint8' if int8 else '',
        '--output_node_names=Identity,Identity_1,Identity_2,Identity_3',
        str(f_pb),
        str(f), ]
    subprocess.run([arg for arg in args if arg], check=True)

    json = Path(f_json).read_text()
    with open(f_json, 'w') as j:  # sort JSON Identity_* in ascending order
        subst = re.sub(
            r'{"outputs": {"Identity.?.?": {"name": "Identity.?.?"}, '
            r'"Identity.?.?": {"name": "Identity.?.?"}, '
            r'"Identity.?.?": {"name": "Identity.?.?"}, '
            r'"Identity.?.?": {"name": "Identity.?.?"}}}', r'{"outputs": {"Identity": {"name": "Identity"}, '
            r'"Identity_1": {"name": "Identity_1"}, '
            r'"Identity_2": {"name": "Identity_2"}, '
            r'"Identity_3": {"name": "Identity_3"}}}', json)
        j.write(subst)
    return f, None


def add_tflite_metadata(file, metadata, num_outputs):
    # Add metadata to *.tflite models per https://www.tensorflow.org/lite/models/convert/metadata
    with contextlib.suppress(ImportError):
        # check_requirements('tflite_support')
        from tflite_support import flatbuffers
        from tflite_support import metadata as _metadata
        from tflite_support import metadata_schema_py_generated as _metadata_fb

        tmp_file = Path('/tmp/meta.txt')
        with open(tmp_file, 'w') as meta_f:
            meta_f.write(str(metadata))

        model_meta = _metadata_fb.ModelMetadataT()
        label_file = _metadata_fb.AssociatedFileT()
        label_file.name = tmp_file.name
        model_meta.associatedFiles = [label_file]

        subgraph = _metadata_fb.SubGraphMetadataT()
        subgraph.inputTensorMetadata = [_metadata_fb.TensorMetadataT()]
        subgraph.outputTensorMetadata = [_metadata_fb.TensorMetadataT()] * num_outputs
        model_meta.subgraphMetadata = [subgraph]

        b = flatbuffers.Builder(0)
        b.Finish(model_meta.Pack(b), _metadata.MetadataPopulator.METADATA_FILE_IDENTIFIER)
        metadata_buf = b.Output()

        populator = _metadata.MetadataPopulator.with_model_file(file)
        populator.load_metadata_buffer(metadata_buf)
        populator.load_associated_files([str(tmp_file)])
        populator.populate()
        tmp_file.unlink()


def pipeline_coreml(model, im, file, names, y, prefix=colorstr('CoreML Pipeline:')):
    # YOLOv5 CoreML pipeline
    import coremltools as ct
    from PIL import Image

    print(f'{prefix} starting pipeline with coremltools {ct.__version__}...')
    batch_size, ch, h, w = list(im.shape)  # BCHW
    t = time.time()

    # YOLOv5 Output shapes
    spec = model.get_spec()
    out0, out1 = iter(spec.description.output)
    if platform.system() == 'Darwin':
        img = Image.new('RGB', (w, h))  # img(192 width, 320 height)
        # img = torch.zeros((*opt.img_size, 3)).numpy()  # img size(320,192,3) iDetection
        out = model.predict({'image': img})
        out0_shape, out1_shape = out[out0.name].shape, out[out1.name].shape
    else:  # linux and windows can not run model.predict(), get sizes from pytorch output y
        s = tuple(y[0].shape)
        out0_shape, out1_shape = (s[1], s[2] - 5), (s[1], 4)  # (3780, 80), (3780, 4)

    # Checks
    nx, ny = spec.description.input[0].type.imageType.width, spec.description.input[0].type.imageType.height
    na, nc = out0_shape
    # na, nc = out0.type.multiArrayType.shape  # number anchors, classes
    assert len(names) == nc, f'{len(names)} names found for nc={nc}'  # check

    # Define output shapes (missing)
    out0.type.multiArrayType.shape[:] = out0_shape  # (3780, 80)
    out1.type.multiArrayType.shape[:] = out1_shape  # (3780, 4)
    # spec.neuralNetwork.preprocessing[0].featureName = '0'

    # Flexible input shapes
    # from coremltools.models.neural_network import flexible_shape_utils
    # s = [] # shapes
    # s.append(flexible_shape_utils.NeuralNetworkImageSize(320, 192))
    # s.append(flexible_shape_utils.NeuralNetworkImageSize(640, 384))  # (height, width)
    # flexible_shape_utils.add_enumerated_image_sizes(spec, feature_name='image', sizes=s)
    # r = flexible_shape_utils.NeuralNetworkImageSizeRange()  # shape ranges
    # r.add_height_range((192, 640))
    # r.add_width_range((192, 640))
    # flexible_shape_utils.update_image_size_range(spec, feature_name='image', size_range=r)

    # Print
    print(spec.description)

    # Model from spec
    model = ct.models.MLModel(spec)

    # 3. Create NMS protobuf
    nms_spec = ct.proto.Model_pb2.Model()
    nms_spec.specificationVersion = 5
    for i in range(2):
        decoder_output = model._spec.description.output[i].SerializeToString()
        nms_spec.description.input.add()
        nms_spec.description.input[i].ParseFromString(decoder_output)
        nms_spec.description.output.add()
        nms_spec.description.output[i].ParseFromString(decoder_output)

    nms_spec.description.output[0].name = 'confidence'
    nms_spec.description.output[1].name = 'coordinates'

    output_sizes = [nc, 4]
    for i in range(2):
        ma_type = nms_spec.description.output[i].type.multiArrayType
        ma_type.shapeRange.sizeRanges.add()
        ma_type.shapeRange.sizeRanges[0].lowerBound = 0
        ma_type.shapeRange.sizeRanges[0].upperBound = -1
        ma_type.shapeRange.sizeRanges.add()
        ma_type.shapeRange.sizeRanges[1].lowerBound = output_sizes[i]
        ma_type.shapeRange.sizeRanges[1].upperBound = output_sizes[i]
        del ma_type.shape[:]

    nms = nms_spec.nonMaximumSuppression
    nms.confidenceInputFeatureName = out0.name  # 1x507x80
    nms.coordinatesInputFeatureName = out1.name  # 1x507x4
    nms.confidenceOutputFeatureName = 'confidence'
    nms.coordinatesOutputFeatureName = 'coordinates'
    nms.iouThresholdInputFeatureName = 'iouThreshold'
    nms.confidenceThresholdInputFeatureName = 'confidenceThreshold'
    nms.iouThreshold = 0.45
    nms.confidenceThreshold = 0.25
    nms.pickTop.perClass = True
    nms.stringClassLabels.vector.extend(names.values())
    nms_model = ct.models.MLModel(nms_spec)

    # 4. Pipeline models together
    pipeline = ct.models.pipeline.Pipeline(input_features=[('image', ct.models.datatypes.Array(3, ny, nx)),
                                                           ('iouThreshold', ct.models.datatypes.Double()),
                                                           ('confidenceThreshold', ct.models.datatypes.Double())],
                                           output_features=['confidence', 'coordinates'])
    pipeline.add_model(model)
    pipeline.add_model(nms_model)

    # Correct datatypes
    pipeline.spec.description.input[0].ParseFromString(model._spec.description.input[0].SerializeToString())
    pipeline.spec.description.output[0].ParseFromString(nms_model._spec.description.output[0].SerializeToString())
    pipeline.spec.description.output[1].ParseFromString(nms_model._spec.description.output[1].SerializeToString())

    # Update metadata
    pipeline.spec.specificationVersion = 5
    pipeline.spec.description.metadata.versionString = 'https://github.com/ultralytics/yolov5'
    pipeline.spec.description.metadata.shortDescription = 'https://github.com/ultralytics/yolov5'
    pipeline.spec.description.metadata.author = '[email protected]'
    pipeline.spec.description.metadata.license = 'https://github.com/ultralytics/yolov5/blob/master/LICENSE'
    pipeline.spec.description.metadata.userDefined.update({
        'classes': ','.join(names.values()),
        'iou_threshold': str(nms.iouThreshold),
        'confidence_threshold': str(nms.confidenceThreshold)})

    # Save the model
    f = file.with_suffix('.mlmodel')  # filename
    model = ct.models.MLModel(pipeline.spec)
    model.input_description['image'] = 'Input image'
    model.input_description['iouThreshold'] = f'(optional) IOU Threshold override (default: {nms.iouThreshold})'
    model.input_description['confidenceThreshold'] = \
        f'(optional) Confidence Threshold override (default: {nms.confidenceThreshold})'
    model.output_description['confidence'] = 'Boxes × Class confidence (see user-defined metadata "classes")'
    model.output_description['coordinates'] = 'Boxes × [x, y, width, height] (relative to image size)'
    model.save(f)  # pipelined
    print(f'{prefix} pipeline success ({time.time() - t:.2f}s), saved as {f} ({file_size(f):.1f} MB)')


@smart_inference_mode()
def run(
        data=ROOT / 'data/coco128.yaml',  # 'dataset.yaml path'
        weights=ROOT / 'yolov5s.pt',  # weights path
        imgsz=(640, 640),  # image (height, width)
        batch_size=1,  # batch size
        device='cpu',  # cuda device, i.e. 0 or 0,1,2,3 or cpu
        include=('torchscript', 'onnx'),  # include formats
        half=False,  # FP16 half-precision export
        inplace=False,  # set YOLOv5 Detect() inplace=True
        keras=False,  # use Keras
        optimize=False,  # TorchScript: optimize for mobile
        int8=False,  # CoreML/TF INT8 quantization
        dynamic=False,  # ONNX/TF/TensorRT: dynamic axes
        simplify=False,  # ONNX: simplify model
        opset=12,  # ONNX: opset version
        verbose=False,  # TensorRT: verbose log
        workspace=4,  # TensorRT: workspace size (GB)
        nms=False,  # TF: add NMS to model
        agnostic_nms=False,  # TF: add agnostic NMS to model
        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
        conf_thres=0.25,  # TF.js NMS: confidence threshold
):
    t = time.time()
    include = [x.lower() for x in include]  # to lowercase
    fmts = tuple(export_formats()['Argument'][1:])  # --include arguments
    flags = [x in include for x in fmts]
    assert sum(flags) == len(include), f'ERROR: Invalid --include {include}, valid --include arguments are {fmts}'
    jit, onnx, xml, engine, coreml, saved_model, pb, tflite, edgetpu, tfjs, paddle = flags  # export booleans
    file = Path(url2file(weights) if str(weights).startswith(('http:/', 'https:/')) else weights)  # PyTorch weights

    # Load PyTorch model
    device = select_device(device)
    if half:
        assert device.type != 'cpu' or coreml, '--half only compatible with GPU export, i.e. use --device 0'
        assert not dynamic, '--half not compatible with --dynamic, i.e. use either --half or --dynamic but not both'
    model = attempt_load(weights, device=device, inplace=True, fuse=True)  # load FP32 model

    # Checks
    imgsz *= 2 if len(imgsz) == 1 else 1  # expand
    if optimize:
        assert device.type == 'cpu', '--optimize not compatible with cuda devices, i.e. use --device cpu'

    # Input
    gs = int(max(model.stride))  # grid size (max stride)
    imgsz = [check_img_size(x, gs) for x in imgsz]  # verify img_size are gs-multiples
    im = torch.zeros(batch_size, 3, *imgsz).to(device)  # image size(1,3,320,192) BCHW iDetection

    # Update model
    model.eval()
    for k, m in model.named_modules():
        if isinstance(m, Detect):
            m.inplace = inplace
            m.dynamic = dynamic
            m.export = True

    for _ in range(2):
        y = model(im)  # dry runs
    if half and not coreml:
        im, model = im.half(), model.half()  # to FP16
    shape = tuple((y[0] if isinstance(y, tuple) else y).shape)  # model output shape
    metadata = {'stride': int(max(model.stride)), 'names': model.names}  # model metadata
    LOGGER.info(f"\n{colorstr('PyTorch:')} starting from {file} with output shape {shape} ({file_size(file):.1f} MB)")

    # Exports
    f = [''] * len(fmts)  # exported filenames
    warnings.filterwarnings(action='ignore', category=torch.jit.TracerWarning)  # suppress TracerWarning
    if jit:  # TorchScript
        f[0], _ = export_torchscript(model, im, file, optimize)
    if engine:  # TensorRT required before ONNX
        f[1], _ = export_engine(model, im, file, half, dynamic, simplify, workspace, verbose)
    if onnx or xml:  # OpenVINO requires ONNX
        f[2], _ = export_onnx(model, im, file, opset, dynamic, simplify)
    if xml:  # OpenVINO
        f[3], _ = export_openvino(file, metadata, half, int8, data)
    if coreml:  # CoreML
        f[4], ct_model = export_coreml(model, im, file, int8, half, nms)
        if nms:
            pipeline_coreml(ct_model, im, file, model.names, y)
    if any((saved_model, pb, tflite, edgetpu, tfjs)):  # TensorFlow formats
        assert not tflite or not tfjs, 'TFLite and TF.js models must be exported separately, please pass only one type.'
        assert not isinstance(model, ClassificationModel), 'ClassificationModel export to TF formats not yet supported.'
        f[5], s_model = export_saved_model(model.cpu(),
                                           im,
                                           file,
                                           dynamic,
                                           tf_nms=nms or agnostic_nms or tfjs,
                                           agnostic_nms=agnostic_nms or tfjs,
                                           topk_per_class=topk_per_class,
                                           topk_all=topk_all,
                                           iou_thres=iou_thres,
                                           conf_thres=conf_thres,
                                           keras=keras)
        if pb or tfjs:  # pb prerequisite to tfjs
            f[6], _ = export_pb(s_model, file)
        if tflite or edgetpu:
            f[7], _ = export_tflite(s_model, im, file, int8 or edgetpu, data=data, nms=nms, agnostic_nms=agnostic_nms)
            if edgetpu:
                f[8], _ = export_edgetpu(file)
            add_tflite_metadata(f[8] or f[7], metadata, num_outputs=len(s_model.outputs))
        if tfjs:
            f[9], _ = export_tfjs(file, int8)
    if paddle:  # PaddlePaddle
        f[10], _ = export_paddle(model, im, file, metadata)

    # Finish
    f = [str(x) for x in f if x]  # filter out '' and None
    if any(f):
        cls, det, seg = (isinstance(model, x) for x in (ClassificationModel, DetectionModel, SegmentationModel))  # type
        det &= not seg  # segmentation models inherit from SegmentationModel(DetectionModel)
        dir = Path('segment' if seg else 'classify' if cls else '')
        h = '--half' if half else ''  # --half FP16 inference arg
        s = '# WARNING ⚠️ ClassificationModel not yet supported for PyTorch Hub AutoShape inference' if cls else \
            '# WARNING ⚠️ SegmentationModel not yet supported for PyTorch Hub AutoShape inference' if seg else ''
        LOGGER.info(f'\nExport complete ({time.time() - t:.1f}s)'
                    f"\nResults saved to {colorstr('bold', file.parent.resolve())}"
                    f"\nDetect:          python {dir / ('detect.py' if det else 'predict.py')} --weights {f[-1]} {h}"
                    f"\nValidate:        python {dir / 'val.py'} --weights {f[-1]} {h}"
                    f"\nPyTorch Hub:     model = torch.hub.load('ultralytics/yolov5', 'custom', '{f[-1]}')  {s}"
                    f'\nVisualize:       https://netron.app')
    return f  # return list of exported files/dirs


def parse_opt(known=False):
    parser = argparse.ArgumentParser()
    parser.add_argument('--data', type=str, default=ROOT / 'data/coco128.yaml', help='dataset.yaml path')
    parser.add_argument('--weights', nargs='+', type=str, default=ROOT / 'yolov5s.pt', help='model.pt path(s)')
    parser.add_argument('--imgsz', '--img', '--img-size', nargs='+', type=int, default=[640, 640], help='image (h, w)')
    parser.add_argument('--batch-size', type=int, default=1, help='batch size')
    parser.add_argument('--device', default='cpu', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
    parser.add_argument('--half', action='store_true', help='FP16 half-precision export')
    parser.add_argument('--inplace', action='store_true', help='set YOLOv5 Detect() inplace=True')
    parser.add_argument('--keras', action='store_true', help='TF: use Keras')
    parser.add_argument('--optimize', action='store_true', help='TorchScript: optimize for mobile')
    parser.add_argument('--int8', action='store_true', help='CoreML/TF/OpenVINO INT8 quantization')
    parser.add_argument('--dynamic', action='store_true', help='ONNX/TF/TensorRT: dynamic axes')
    parser.add_argument('--simplify', action='store_true', help='ONNX: simplify model')
    parser.add_argument('--opset', type=int, default=17, 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('--nms', action='store_true', help='TF: add NMS to model')
    parser.add_argument('--agnostic-nms', action='store_true', help='TF: add agnostic NMS to model')
    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'],
        help='torchscript, onnx, openvino, engine, coreml, saved_model, pb, tflite, edgetpu, tfjs, paddle')
    opt = parser.parse_known_args()[0] if known else parser.parse_args()
    print_args(vars(opt))
    return opt


def main(opt):
    for opt.weights in (opt.weights if isinstance(opt.weights, list) else [opt.weights]):
        run(**vars(opt))


if __name__ == '__main__':
    opt = parse_opt()
    main(opt)