TrojanXu / yolov5-tensorrt

A tensorrt implementation of yolov5: https://github.com/ultralytics/yolov5
Apache License 2.0
190 stars 46 forks source link

how to build on Jetson Xavier? #5

Closed batrlatom closed 4 years ago

batrlatom commented 4 years ago

Hi, I tried it with Xavier ( aarch64) , but I am unable to install

onnx-simplify .

How can I make it run on Jetson Xavier?

Trying to go without simplifier got an error:

` %549 : Tensor = onnx::Shape(%548) %550 : Tensor = onnx::Constant[value={0}]() %551 : Long() = onnx::Gather[axis=0](%549, %550) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:28:0 %552 : Tensor = onnx::Shape(%548) %553 : Tensor = onnx::Constant[value={2}]() %554 : Long() = onnx::Gather[axis=0](%552, %553) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:28:0 %555 : Tensor = onnx::Shape(%548) %556 : Tensor = onnx::Constant[value={3}]() %557 : Long() = onnx::Gather[axis=0](%555, %556) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:28:0 %560 : Tensor = onnx::Unsqueezeaxes=[0] %563 : Tensor = onnx::Unsqueezeaxes=[0] %564 : Tensor = onnx::Unsqueezeaxes=[0] %565 : Tensor = onnx::Concat[axis=0](%560, %856, %857, %563, %564) %566 : Float(1, 3, 85, 80, 64) = onnx::Reshape(%548, %565) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:29:0 %567 : Float(1, 3, 80, 64, 85) = onnx::Transposeperm=[0, 1, 3, 4, 2] # /home/pekat/Devel/yolov5-tensorrt/yolo.py:29:0 %568 : Tensor = onnx::Unsqueezeaxes=[0] %569 : Tensor = onnx::ConstantOfShapevalue={1} %570 : Tensor = onnx::NonZero(%569) %571 : Tensor = onnx::Transposeperm=[1, 0] %572 : Tensor = onnx::Squeezeaxes=[1] %573 : Long(80) = onnx::Castto=7 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:47:0 %574 : Tensor = onnx::Unsqueezeaxes=[0] %575 : Tensor = onnx::ConstantOfShapevalue={1} %576 : Tensor = onnx::NonZero(%575) %577 : Tensor = onnx::Transposeperm=[1, 0] %578 : Tensor = onnx::Squeezeaxes=[1] %579 : Long(64) = onnx::Castto=7 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:47:0 %580 : Tensor = onnx::Constant[value={-1}]() %581 : Tensor = onnx::Reshape(%573, %580) %582 : Tensor = onnx::Constant[value={-1}]() %583 : Tensor = onnx::Reshape(%579, %582) %584 : Tensor = onnx::Shape(%581) %585 : Tensor = onnx::Shape(%583) %586 : Tensor = onnx::Concat[axis=0](%584, %585) %587 : Tensor = onnx::Constant[value={1}]() %588 : Tensor = onnx::Concat[axis=0](%584, %587) %589 : Tensor = onnx::Reshape(%581, %588) %590 : Tensor = onnx::Expand(%589, %586) %591 : Tensor = onnx::Constant[value={1}]() %592 : Tensor = onnx::Concat[axis=0](%591, %585) %593 : Tensor = onnx::Reshape(%583, %592) %594 : Tensor = onnx::Expand(%593, %586) %595 : Tensor = onnx::Unsqueezeaxes=[2] %596 : Tensor = onnx::Unsqueezeaxes=[2] %597 : Long(80, 64, 2) = onnx::Concat[axis=2](%595, %596) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:48:0 %603 : Tensor = onnx::Unsqueezeaxes=[0] %604 : Tensor = onnx::Unsqueezeaxes=[0] %606 : Tensor = onnx::Concat[axis=0](%858, %859, %603, %604, %860) %607 : Long(1, 1, 80, 64, 2) = onnx::Reshape(%597, %606) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:48:0 %608 : Float(1, 1, 80, 64, 2) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:48:0 %609 : Float(1, 1, 80, 64, 2) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:33:0 %610 : Float(1, 3, 80, 64, 85) = onnx::Sigmoid(%567) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:35:0 %611 : Tensor = onnx::Constant[value={4}]() %612 : Tensor = onnx::Constant[value={0}]() %613 : Tensor = onnx::Constant[value={2}]() %614 : Tensor = onnx::Constant[value={1}]() %615 : Float(1, 3, 80, 64, 2) = onnx::Slice(%610, %612, %613, %611, %614) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:38:0 %616 : Float() = onnx::Constant[value={2}]() %617 : Float(1, 3, 80, 64, 2) = onnx::Mul(%615, %616) %618 : Float() = onnx::Constant[value={0.5}]() %619 : Float(1, 3, 80, 64, 2) = onnx::Sub(%617, %618) %620 : Float(1, 1, 80, 64, 2) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:38:0 %621 : Float(1, 3, 80, 64, 2) = onnx::Add(%619, %620) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:38:0 %622 : Float() = onnx::Constant[value={8}]() %623 : Float(1, 3, 80, 64, 2) = onnx::Mul(%621, %622) %624 : Tensor = onnx::Constant[value={4}]() %625 : Tensor = onnx::Constant[value={2}]() %626 : Tensor = onnx::Constant[value={4}]() %627 : Tensor = onnx::Constant[value={1}]() %628 : Float(1, 3, 80, 64, 2) = onnx::Slice(%610, %625, %626, %624, %627) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:39:0 %629 : Float() = onnx::Constant[value={2}]() %630 : Float(1, 3, 80, 64, 2) = onnx::Mul(%628, %629) %631 : Float() = onnx::Constant[value={2}]() %632 : Float(1, 3, 80, 64, 2) = onnx::Pow(%630, %631) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:39:0 %633 : Tensor = onnx::Constant[value={0}]() %634 : Float(1, 3, 1, 1, 2) = onnx::Gather[axis=0](%model.22.anchor_grid, %633) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:39:0 %635 : Float(1, 3, 80, 64, 2) = onnx::Mul(%632, %634) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:39:0 %636 : Float(1, 3, 80, 64, 2) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:40:0 %637 : Tensor = onnx::Constant[value={4}]() %638 : Tensor = onnx::Constant[value={4}]() %639 : Tensor = onnx::Constant[value={9223372036854775807}]() %640 : Tensor = onnx::Constant[value={1}]() %641 : Float(1, 3, 80, 64, 81) = onnx::Slice(%610, %638, %639, %637, %640) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:40:0 %642 : Float(1, 3, 80, 64, 81) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:40:0 %643 : Float(1, 3, 80, 64, 85) = onnx::Concat[axis=-1](%623, %636, %642) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:40:0 %646 : Tensor = onnx::Unsqueezeaxes=[0] %649 : Tensor = onnx::Concat[axis=0](%646, %861, %862) %650 : Float(1, 15360, 85) = onnx::Reshape(%643, %649) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:41:0 %651 : Tensor = onnx::Shape(%524) %652 : Tensor = onnx::Constant[value={0}]() %653 : Long() = onnx::Gather[axis=0](%651, %652) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:28:0 %654 : Tensor = onnx::Shape(%524) %655 : Tensor = onnx::Constant[value={2}]() %656 : Long() = onnx::Gather[axis=0](%654, %655) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:28:0 %657 : Tensor = onnx::Shape(%524) %658 : Tensor = onnx::Constant[value={3}]() %659 : Long() = onnx::Gather[axis=0](%657, %658) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:28:0 %662 : Tensor = onnx::Unsqueezeaxes=[0] %665 : Tensor = onnx::Unsqueezeaxes=[0] %666 : Tensor = onnx::Unsqueezeaxes=[0] %667 : Tensor = onnx::Concat[axis=0](%662, %863, %864, %665, %666) %668 : Float(1, 3, 85, 40, 32) = onnx::Reshape(%524, %667) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:29:0 %669 : Float(1, 3, 40, 32, 85) = onnx::Transposeperm=[0, 1, 3, 4, 2] # /home/pekat/Devel/yolov5-tensorrt/yolo.py:29:0 %670 : Tensor = onnx::Unsqueezeaxes=[0] %671 : Tensor = onnx::ConstantOfShapevalue={1} %672 : Tensor = onnx::NonZero(%671) %673 : Tensor = onnx::Transposeperm=[1, 0] %674 : Tensor = onnx::Squeezeaxes=[1] %675 : Long(40) = onnx::Castto=7 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:47:0 %676 : Tensor = onnx::Unsqueezeaxes=[0] %677 : Tensor = onnx::ConstantOfShapevalue={1} %678 : Tensor = onnx::NonZero(%677) %679 : Tensor = onnx::Transposeperm=[1, 0] %680 : Tensor = onnx::Squeezeaxes=[1] %681 : Long(32) = onnx::Castto=7 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:47:0 %682 : Tensor = onnx::Constant[value={-1}]() %683 : Tensor = onnx::Reshape(%675, %682) %684 : Tensor = onnx::Constant[value={-1}]() %685 : Tensor = onnx::Reshape(%681, %684) %686 : Tensor = onnx::Shape(%683) %687 : Tensor = onnx::Shape(%685) %688 : Tensor = onnx::Concat[axis=0](%686, %687) %689 : Tensor = onnx::Constant[value={1}]() %690 : Tensor = onnx::Concat[axis=0](%686, %689) %691 : Tensor = onnx::Reshape(%683, %690) %692 : Tensor = onnx::Expand(%691, %688) %693 : Tensor = onnx::Constant[value={1}]() %694 : Tensor = onnx::Concat[axis=0](%693, %687) %695 : Tensor = onnx::Reshape(%685, %694) %696 : Tensor = onnx::Expand(%695, %688) %697 : Tensor = onnx::Unsqueezeaxes=[2] %698 : Tensor = onnx::Unsqueezeaxes=[2] %699 : Long(40, 32, 2) = onnx::Concat[axis=2](%697, %698) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:48:0 %705 : Tensor = onnx::Unsqueezeaxes=[0] %706 : Tensor = onnx::Unsqueezeaxes=[0] %708 : Tensor = onnx::Concat[axis=0](%865, %866, %705, %706, %867) %709 : Long(1, 1, 40, 32, 2) = onnx::Reshape(%699, %708) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:48:0 %710 : Float(1, 1, 40, 32, 2) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:48:0 %711 : Float(1, 1, 40, 32, 2) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:33:0 %712 : Float(1, 3, 40, 32, 85) = onnx::Sigmoid(%669) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:35:0 %713 : Tensor = onnx::Constant[value={4}]() %714 : Tensor = onnx::Constant[value={0}]() %715 : Tensor = onnx::Constant[value={2}]() %716 : Tensor = onnx::Constant[value={1}]() %717 : Float(1, 3, 40, 32, 2) = onnx::Slice(%712, %714, %715, %713, %716) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:38:0 %718 : Float() = onnx::Constant[value={2}]() %719 : Float(1, 3, 40, 32, 2) = onnx::Mul(%717, %718) %720 : Float() = onnx::Constant[value={0.5}]() %721 : Float(1, 3, 40, 32, 2) = onnx::Sub(%719, %720) %722 : Float(1, 1, 40, 32, 2) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:38:0 %723 : Float(1, 3, 40, 32, 2) = onnx::Add(%721, %722) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:38:0 %724 : Float() = onnx::Constant[value={16}]() %725 : Float(1, 3, 40, 32, 2) = onnx::Mul(%723, %724) %726 : Tensor = onnx::Constant[value={4}]() %727 : Tensor = onnx::Constant[value={2}]() %728 : Tensor = onnx::Constant[value={4}]() %729 : Tensor = onnx::Constant[value={1}]() %730 : Float(1, 3, 40, 32, 2) = onnx::Slice(%712, %727, %728, %726, %729) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:39:0 %731 : Float() = onnx::Constant[value={2}]() %732 : Float(1, 3, 40, 32, 2) = onnx::Mul(%730, %731) %733 : Float() = onnx::Constant[value={2}]() %734 : Float(1, 3, 40, 32, 2) = onnx::Pow(%732, %733) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:39:0 %735 : Tensor = onnx::Constant[value={1}]() %736 : Float(1, 3, 1, 1, 2) = onnx::Gather[axis=0](%model.22.anchor_grid, %735) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:39:0 %737 : Float(1, 3, 40, 32, 2) = onnx::Mul(%734, %736) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:39:0 %738 : Float(1, 3, 40, 32, 2) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:40:0 %739 : Tensor = onnx::Constant[value={4}]() %740 : Tensor = onnx::Constant[value={4}]() %741 : Tensor = onnx::Constant[value={9223372036854775807}]() %742 : Tensor = onnx::Constant[value={1}]() %743 : Float(1, 3, 40, 32, 81) = onnx::Slice(%712, %740, %741, %739, %742) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:40:0 %744 : Float(1, 3, 40, 32, 81) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:40:0 %745 : Float(1, 3, 40, 32, 85) = onnx::Concat[axis=-1](%725, %738, %744) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:40:0 %748 : Tensor = onnx::Unsqueezeaxes=[0] %751 : Tensor = onnx::Concat[axis=0](%748, %868, %869) %752 : Float(1, 3840, 85) = onnx::Reshape(%745, %751) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:41:0 %753 : Tensor = onnx::Shape(%500) %754 : Tensor = onnx::Constant[value={0}]() %755 : Long() = onnx::Gather[axis=0](%753, %754) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:28:0 %756 : Tensor = onnx::Shape(%500) %757 : Tensor = onnx::Constant[value={2}]() %758 : Long() = onnx::Gather[axis=0](%756, %757) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:28:0 %759 : Tensor = onnx::Shape(%500) %760 : Tensor = onnx::Constant[value={3}]() %761 : Long() = onnx::Gather[axis=0](%759, %760) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:28:0 %764 : Tensor = onnx::Unsqueezeaxes=[0] %767 : Tensor = onnx::Unsqueezeaxes=[0] %768 : Tensor = onnx::Unsqueezeaxes=[0] %769 : Tensor = onnx::Concat[axis=0](%764, %870, %871, %767, %768) %770 : Float(1, 3, 85, 20, 16) = onnx::Reshape(%500, %769) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:29:0 %771 : Float(1, 3, 20, 16, 85) = onnx::Transposeperm=[0, 1, 3, 4, 2] # /home/pekat/Devel/yolov5-tensorrt/yolo.py:29:0 %772 : Tensor = onnx::Unsqueezeaxes=[0] %773 : Tensor = onnx::ConstantOfShapevalue={1} %774 : Tensor = onnx::NonZero(%773) %775 : Tensor = onnx::Transposeperm=[1, 0] %776 : Tensor = onnx::Squeezeaxes=[1] %777 : Long(20) = onnx::Castto=7 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:47:0 %778 : Tensor = onnx::Unsqueezeaxes=[0] %779 : Tensor = onnx::ConstantOfShapevalue={1} %780 : Tensor = onnx::NonZero(%779) %781 : Tensor = onnx::Transposeperm=[1, 0] %782 : Tensor = onnx::Squeezeaxes=[1] %783 : Long(16) = onnx::Castto=7 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:47:0 %784 : Tensor = onnx::Constant[value={-1}]() %785 : Tensor = onnx::Reshape(%777, %784) %786 : Tensor = onnx::Constant[value={-1}]() %787 : Tensor = onnx::Reshape(%783, %786) %788 : Tensor = onnx::Shape(%785) %789 : Tensor = onnx::Shape(%787) %790 : Tensor = onnx::Concat[axis=0](%788, %789) %791 : Tensor = onnx::Constant[value={1}]() %792 : Tensor = onnx::Concat[axis=0](%788, %791) %793 : Tensor = onnx::Reshape(%785, %792) %794 : Tensor = onnx::Expand(%793, %790) %795 : Tensor = onnx::Constant[value={1}]() %796 : Tensor = onnx::Concat[axis=0](%795, %789) %797 : Tensor = onnx::Reshape(%787, %796) %798 : Tensor = onnx::Expand(%797, %790) %799 : Tensor = onnx::Unsqueezeaxes=[2] %800 : Tensor = onnx::Unsqueezeaxes=[2] %801 : Long(20, 16, 2) = onnx::Concat[axis=2](%799, %800) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:48:0 %807 : Tensor = onnx::Unsqueezeaxes=[0] %808 : Tensor = onnx::Unsqueezeaxes=[0] %810 : Tensor = onnx::Concat[axis=0](%872, %873, %807, %808, %874) %811 : Long(1, 1, 20, 16, 2) = onnx::Reshape(%801, %810) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:48:0 %812 : Float(1, 1, 20, 16, 2) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:48:0 %813 : Float(1, 1, 20, 16, 2) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:33:0 %814 : Float(1, 3, 20, 16, 85) = onnx::Sigmoid(%771) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:35:0 %815 : Tensor = onnx::Constant[value={4}]() %816 : Tensor = onnx::Constant[value={0}]() %817 : Tensor = onnx::Constant[value={2}]() %818 : Tensor = onnx::Constant[value={1}]() %819 : Float(1, 3, 20, 16, 2) = onnx::Slice(%814, %816, %817, %815, %818) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:38:0 %820 : Float() = onnx::Constant[value={2}]() %821 : Float(1, 3, 20, 16, 2) = onnx::Mul(%819, %820) %822 : Float() = onnx::Constant[value={0.5}]() %823 : Float(1, 3, 20, 16, 2) = onnx::Sub(%821, %822) %824 : Float(1, 1, 20, 16, 2) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:38:0 %825 : Float(1, 3, 20, 16, 2) = onnx::Add(%823, %824) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:38:0 %826 : Float() = onnx::Constant[value={32}]() %827 : Float(1, 3, 20, 16, 2) = onnx::Mul(%825, %826) %828 : Tensor = onnx::Constant[value={4}]() %829 : Tensor = onnx::Constant[value={2}]() %830 : Tensor = onnx::Constant[value={4}]() %831 : Tensor = onnx::Constant[value={1}]() %832 : Float(1, 3, 20, 16, 2) = onnx::Slice(%814, %829, %830, %828, %831) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:39:0 %833 : Float() = onnx::Constant[value={2}]() %834 : Float(1, 3, 20, 16, 2) = onnx::Mul(%832, %833) %835 : Float() = onnx::Constant[value={2}]() %836 : Float(1, 3, 20, 16, 2) = onnx::Pow(%834, %835) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:39:0 %837 : Tensor = onnx::Constant[value={2}]() %838 : Float(1, 3, 1, 1, 2) = onnx::Gather[axis=0](%model.22.anchor_grid, %837) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:39:0 %839 : Float(1, 3, 20, 16, 2) = onnx::Mul(%836, %838) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:39:0 %840 : Float(1, 3, 20, 16, 2) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:40:0 %841 : Tensor = onnx::Constant[value={4}]() %842 : Tensor = onnx::Constant[value={4}]() %843 : Tensor = onnx::Constant[value={9223372036854775807}]() %844 : Tensor = onnx::Constant[value={1}]() %845 : Float(1, 3, 20, 16, 81) = onnx::Slice(%814, %842, %843, %841, %844) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:40:0 %846 : Float(1, 3, 20, 16, 81) = onnx::Castto=1 # /home/pekat/Devel/yolov5-tensorrt/yolo.py:40:0 %847 : Float(1, 3, 20, 16, 85) = onnx::Concat[axis=-1](%827, %840, %846) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:40:0 %850 : Tensor = onnx::Unsqueezeaxes=[0] %853 : Tensor = onnx::Concat[axis=0](%850, %875, %876) %854 : Float(1, 960, 85) = onnx::Reshape(%847, %853) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:41:0 %model/22 : Float(1, 20160, 85) = onnx::Concat[axis=1](%650, %752, %854) # /home/pekat/Devel/yolov5-tensorrt/yolo.py:43:0 return (%model/22)

[TensorRT] WARNING: onnx2trt_utils.cpp:217: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32. [TensorRT] WARNING: onnx2trt_utils.cpp:243: One or more weights outside the range of INT32 was clamped [TensorRT] WARNING: onnx2trt_utils.cpp:243: One or more weights outside the range of INT32 was clamped [TensorRT] WARNING: onnx2trt_utils.cpp:243: One or more weights outside the range of INT32 was clamped [TensorRT] WARNING: onnx2trt_utils.cpp:243: One or more weights outside the range of INT32 was clamped [TensorRT] WARNING: onnx2trt_utils.cpp:243: One or more weights outside the range of INT32 was clamped [TensorRT] WARNING: onnx2trt_utils.cpp:243: One or more weights outside the range of INT32 was clamped [TensorRT] WARNING: onnx2trt_utils.cpp:243: One or more weights outside the range of INT32 was clamped [TensorRT] WARNING: onnx2trt_utils.cpp:243: One or more weights outside the range of INT32 was clamped [TensorRT] WARNING: ModelImporter.cpp:135: No importer registered for op: NonZero. Attempting to import as plugin. [TensorRT] ERROR: INVALID_ARGUMENT: getPluginCreator could not find plugin NonZero version 001 ERROR: Failed to parse the ONNX file. In node -1 (importFallbackPluginImporter): UNSUPPORTED_NODE: Assertion failed: creator && "Plugin not found" Traceback (most recent call last): File "main.py", line 212, in trt_result = profile_trt(build_engine(onnx_path, using_half), batch_size, 10, 100) File "main.py", line 118, in profile_trt assert(engine is not None) AssertionError `

TrojanXu commented 4 years ago

I don't have jetson environment. I think you can run onnx-simpifier manually on a x86 server.

batrlatom commented 4 years ago

You mean like simplify model on x86 and then use this model on xavier. right? I am going to try. Thanks!

batrlatom commented 4 years ago

It works!!!! Thanks

zhucheng725 commented 4 years ago

It works!!!! Thanks

You can git clone https://github.com/daquexian/onnx-simplifier and python3 setup.py install after installing the onnxruntime and onnx.

I have builded these on Jetson TX2

rwin94 commented 4 years ago

@batrlatom what king of performance are you getting on jetson board if you dont mind me asking?