How to replace upsample with deconvolution?

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

Hello, everyone

I want to implement YoloV3 on my TX2 by using TensorRT.

I have already transfer Darknet model to Caffe model and implement YoloV2 by TensorRT.

I found TensorRT is not support upsample layer and I have referenced some information that it could replace upsample layer with deconvolution layer.

Is there any hint to me how to do it?

Thanks.

2

Hi,

related post on jetson forum:

[i]We have caffe-to-TensorRT parser on Jetson.

We are not sure if each layer of YOLO3 is supported by TensorRT and parser.
Please check it with the information here:
https://docs.nvidia.com/deeplearning/sdk/tensorrt-developer-guide/index.html#layers

If there is a layer doesn’t support by TensorRT yet, you can write your implement with our plugin API.
Here is an example for your reference: https://github.com/AastaNV/Face-Recognition[/i]