This code is sample to implement caffe-ssd.
This sample implements a fast face detector based on caffe-ssd framework.
This model can run more than 250fps on gtx1080ti!
The code has been tested by myself, it can help you learn about TensorRT API and ssd fast!
*1.TensorRT 4.0.1
*2.Cuda8.0 / Cuda9.0 and Cudnn 7.1
*3.OpenCV
Scale layer has been supported officially, but you must provide both scale param and bias param!!!
If you just provide scale params, it won't work!
Eltwise layer has been supported officially too, but you can only provide two blobs as input!!!