Implement VGG-SSD tensor layer by using TensorRT API
- This project is based on qfgaohao/pytorch-ssd and tjuskyzhang/mobilenetv1-ssd-tensorrt. This project tested on jetson agx xavier using TensorRt 18.104.22.168 Cuda10.2 and costs about 1240ms on the contrasts of 2.8s by using python.
Due to time, I did not complete the training of the model,you can train your own model by your datasets and generates wts weights using pthTowts.py
build and run
mkdir build cd build cmake .. make
// Serialize the model and generate ssd_mobilenet.engine
// Deserialize and generate the detection results test.jpg and so on.
./vgg-tensorrt -d ../samples