MobileNet-SSD-TensorRT

To accelerate mobileNet-ssd with tensorRT

TensorRT-Mobilenet-SSD can run 50fps on jetson tx2

Requierments:

1.tensorRT4

2.cudnn7

3.opencv

Run:

cmake .make
./build/bin/mobileNet

Reference:

https://github.com/saikumarGadde/tensorrt-ssd-easy

https://github.com/chuanqi305/MobileNet-SSD

I replaced depthwise with group_conv,because group_conv has been optimized in cudnn7

I retrianed mobileNet-SSD,my number of classfication is 5

TODO:

•  To save serialized model

•  To solve the bug of getting different result with same input

•  The bottleneck of time cost lies in the decoding of pictures. "imread" cost too much ,to resolve it.

•  To modify the architecture, decrease the time cost

If want to decrease the time cost of "imread",you could rebuild OpenCV[https://github.com/jetsonhacks/buildOpenCVTX2]

Added producer-consumer