Rookie_Chris

Did you finish the two steps（pre-processing and post-processing）？ @tengerye

You should implement Mobilenet-SSD by referring to my code, and it is most important to implement the depth-wise convolution layer with Iplugin API.

我记得好像是300维

1. You should see some tensor samples, for example the samplePlugin folder， the sampleFasterRCNN folder. 2. If you want to implement the plugin layer, you have to do CUDA programming.

2-6 months

I hope my code will help you learn and understand the TensorRT API better, instead of just using my code.So， firstly， you should know about TensorRT ， and we can...

Yes

ssd+vgg16： inference time 80-100ms ssd+Mobilenetv1： inference time 15-20ms

Hi, In my jetson-tx2, I got twice the acceleration by using TensorRT on the ssd-vgg16 architecture. Maybe you have a problem with the implementation of your TensorRT code.

You can refer to this code.(https://github.com/AastaNV/Face-Recognition)