https://github.com/kunle12/dn_object_detect

Multiclass object detection ROS node based on YOLO/Darknet
https://github.com/kunle12/dn_object_detect

Last synced: 3 months ago
JSON representation

Multiclass object detection ROS node based on YOLO/Darknet

• Host: GitHub
• URL: https://github.com/kunle12/dn_object_detect
• Owner: kunle12
• License: bsd-2-clause
• Created: 2016-05-12T14:35:57.000Z (about 8 years ago)
• Default Branch: master
• Last Pushed: 2019-02-04T03:29:02.000Z (over 5 years ago)
• Last Synced: 2024-01-18T13:08:36.741Z (5 months ago)
• Language: C++
• Homepage:
• Size: 33.2 KB
• Stars: 39
• Watchers: 10
• Forks: 24
• Open Issues: 4
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Lists

• awesome-self-driving-cars - YOLO 2 in ROS

README

        
# Multi-class object detection with darknet/yolo

## Introduction

This is a ROS node for muli-class object detection using darknet/yolo Deep Neural Network (DNN).

## Compilation

Fully check out this repository with submodule ```darknet```. Compile the ```darknet``` library with the following commands:

```

cd darknet;mkdir build;cd build

cmake -DCMAKE_INSTALL_PREFIX= ..

make install

```

**NOTE** If you are compiling darknet submodule with OpenCV 3.x, you need to add the following lines in ```opencv2/core/types_c.h``` after ```#include "opencv2/core/cvdef.h"```:

```

#ifndef __cplusplus

#include "opencv2/core/fast_math.hpp"

#endif

```

Go back to the catkin workspace base directory and compile the ROS node with

```

catkin_make dn_object_detect

```

## Running Node

**NOTE** You need to download the model/weights file separately.

```roslaunch dn_object_detect objdetect.launch```

The node will publish the following two topics

```/dn_object_detect/detected_objects```  detected object list.

```/dn_object_detect/debug_view``` debugging image stream.