Apollo 10.0
自动驾驶开放平台
README

perception-camera-detection-multi-stage

Introduction

The camera 2D object detection module is a multitask model developed based on early yolo, which can simultaneously output dozens of dimensional information such as 2d ,3d, and vehicle turn signals. This module completes operations such as image data preprocessing, detection, result post-processing and so on.

Directory Structure

├── camera_detection_multi_stage // camera detect 2d module
├── conf // module configuration files
├── dag // dag files
├── data // model params
├── detector // main part for detector
│ ├── yolo // YoloObstacleDetector
│ └── ...
├── interface // function interface folder
├── proto // proto files
├── camera_detection_multi_stage_component.cc // component interface
├── camera_detection_multi_stage_component.h
├── cyberfile.xml // package management profile
├── README.md
└── BUILD

Modules

CameraDetectionMultiStageComponent

apollo::perception::camera::CameraDetectionMultiStageComponent

Input

Name Type Description Input channal
msg apollo::drivers::Image camera sensor image /apollo/sensor/camera/front_6mm/image

>Note: Enter the data type defined by proto. The default trigger camera channel is /apollo/sensor/camera/front_6mm/image. The detailed input channel information is in modules/perception/camera_detection_multi_stage/dag/camera_detection_multi_stage_yolox3d.dag file.

Output

Name Type Description Output channal
frame apollo::perception::onboard::CameraFrame camera frame message /perception/inner/Detection

>Note: The output channel is structure type data. The message is defined in the modules/perception/common/onboard/inner_component_messages/camera_detection_component_messages.h file. The output channel message data can be subscribed by components in the same process. The detailed output channel information is in modules/perception/camera_detection_multi_stage/conf/camera_detection_multi_stage_yolox3d_config.pb.txt file.

How to Launch

cyber_launch start modules/perception/launch/perception_camera_multi_stage.launch

Reference

  1. YOLO: Real-Time Object Detection
  2. 3D Bounding Box Estimation Using Deep Learning and Geometry

文档意见反馈

如果您在使用文档的过程中,遇到任何问题,请到我们在【开发者社区】建立的 反馈意见收集问答页面,反馈相关的问题。我们会根据反馈意见对文档进行迭代优化。