Skip to content

faedtodd/Tensorrt-Yolov3-tiny

Repository files navigation

Tensorrt-Yolov3-tiny

darknet.weights --> onnx.onnx --> tensorrt.trt

sample from TensorRT-5.x.x.x/samples/python/yolov3_onnx/

Environment

device:
nvidia tx2

environment:
ubuntu18.04
tensorrt5.0.6.3
cuda10.0
cudnn7.3.1

How to use

set input size and batch size

set labels: line14,line19|data_processing.py
set output_layers: line723|yolov3_to_onnx.py
set batch_size: batch|xxx.cfg, line62|yolov3_to_onnx.py, line140|onnx_to_tensorrt.py
set input_size: input|xxx.cfg, line63|yolov3_to_onnx.py, line139|onnx_to_tensorrt.py
set mode: line141|onnx_to_tensorrt.py

performance on Tx2 with turn fp16 mode on:

input      batch size    inference time/ms
416x416        1                9.5
               16               8.75
               32               8.5
608x608        1                20
               16               18.1
               32               18

About

darknet2onnx2tensorrt

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages