From f5bbd1031bad89f240450b322ae6bcc6180158fd Mon Sep 17 00:00:00 2001 From: Guanghua Yu <742925032@qq.com> Date: Fri, 26 Aug 2022 11:48:48 +0800 Subject: [PATCH] update yolo tensorrt docs --- .../pytorch_yolo_series/TensorRT/README.md | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/example/auto_compression/pytorch_yolo_series/TensorRT/README.md b/example/auto_compression/pytorch_yolo_series/TensorRT/README.md index a755cb4d..82b5753f 100644 --- a/example/auto_compression/pytorch_yolo_series/TensorRT/README.md +++ b/example/auto_compression/pytorch_yolo_series/TensorRT/README.md @@ -1,8 +1,12 @@ # TensorRT Python预测 +### 环境要求 + +- TensorRT 8.0以上 + ### 验证COCO mAP --FP16 +- FP16 ```shell python trt_eval.py --onnx_model_file=yolov7_tiny_quant_onnx/yolov7-tiny.onnx \ --precision_mode=fp16 \ @@ -38,4 +42,4 @@ python trt_eval.py --onnx_model_file=yolov7_tiny_quant_onnx/yolov7_tiny_quant.on ### FAQ -- 测试内存和显存占用时,首次运行会将ONNX模型转换成TRT模型,耗时不准确,再此运行trt_eval.py可获取真实的内存和显存占用。 +- 测试内存和显存占用时,首次运行会将ONNX模型转换成TRT模型,耗时不准确,再次运行trt_eval.py可获取真实的内存和显存占用。 -- GitLab