Bohrhh / trt_inference

Tensorrt Inference Sample Code

Requirements

• tensorrt==7.x
• opencv==3.4.x
• yaml-cpp
• tensorrt_plugin

Directory

├── configs/
│   ├── camera/ camera calibration files
│   ├── models/ model config files
│
├── common/  # nvidia common utils
│
├── include/
│
├── src/
│   ├── stereo_camera.cpp
│   ├── preprocess.cpp
│   ├── base_model.cpp
│   ├── main.cpp
│   ├── models/
│   │   ├── feature_extraction.cpp
│   │   ├── instance_seg.cpp
│   │   ├── mono_depth.cpp
│   │   ├── stereo_depth.cpp
│   │   ├── yolov5.cpp

说明：models 文件夹下是具体的模型类别，需要继承 base_model 并且实现自己的可视化方法。

Usage

mkdir build && cd build
cmake ..
make

cd ..
build/inference configs/fastFeat.yaml
# settings of camera and model are in yaml file