zh/integrations/tensorrt/ #9868
Replies: 6 comments 8 replies
-
请问在那儿可以看到 |
Beta Was this translation helpful? Give feedback.
-
(yolov8) PS E:\ultralytics> yolo export model=./weights/yolov8n.pt format=engine PyTorch: starting from 'weights\yolov8n.pt' with input shape (1, 3, 640, 640) BCHW and output shape(s) (1, 84, 8400) (6.2 MB) ONNX: starting export with onnx 1.15.0 opset 17... |
Beta Was this translation helpful? Give feedback.
-
when I use official demo to export into TensorRT, it gave me a feedback on the following error: AttributeError: 'tensorrt_bindings.tensorrt.IBuilderConfig' object has no attribute 'max_workspace_size' the system is Windows 11, and the CUDA version is 12.1, the TensorRT version is 10.1 |
Beta Was this translation helpful? Give feedback.
-
您好,我在使用yolov8.pt权重转换成engine格式进行测试时,发现engine还没有原先的.pt快,这是两者相比较的速度 best-seg.engine 其它参数都是一样的 model = YOLO(r"best-seg.pt") # load a custom trained model Export the modelmodel.export( 这是推理代码:
我的疑惑是也调用了GPU进行加速推理,然后推理结果反而比原来的慢,我不清楚是什么原因,请解答我的疑惑,谢谢。 |
Beta Was this translation helpful? Give feedback.
-
AttributeError: 'tensorrt_bindings.tensorrt.ICudaEngine' object has no attribute 'num_bindings' Load the exported TensorRT modeltensorrt_model = YOLO("yolov8n.engine") Run inferenceresults = tensorrt_model("https://ultralytics.com/images/bus.jpg") |
Beta Was this translation helpful? Give feedback.
-
Ultralytics是否支持对RT-DETR的TensorRT模型进行export以及predict? |
Beta Was this translation helpful? Give feedback.
-
zh/integrations/tensorrt/
探索将Ultralytics YOLOv8 模型导出为TensorRT 格式的强大功能和灵活性,以提高在英伟达™(NVIDIA®)图形处理器上的性能和效率。
https://docs.ultralytics.com/zh/integrations/tensorrt/
Beta Was this translation helpful? Give feedback.
All reactions