The recall rate of tiny-yolov3 with trt5 is lower than darknet

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

Description

The tiny-yolov3 is trained by darknet,and I get the tensorrt engine by tensorrt c++ api, the precision is kfloat.
I test the tensorrt engine of tiny-yolov3 with 500 pictures，the engine can find 1864 objects.
However, I can get 2121 objects by runing darknet with the same picture.
How can I optimizate the engine？
Thanks.

Environment

Jetson TX2 R32.1

TensorRT Version: 5

3

Hi @qgq1991,

We recommend you to try with latest TensorRT version.

Thank you.

4

I have got almost the same result with the latest TensorRT version.

5

Hi @qgq1991,

Could you please share more details, which TRT version you tried and issue repro minimal script, model to try from our end for better assistance.

Thank you.