This repository contains the dataset collection scripts for R&D project for depth completion and the title of the project is "Fusing 3d Range Measures with high resolution RGB data for obstacle detection and avoidance"
The code was tested in ROS-Kinetic on Ubuntu 16.04.
- Data collection scripts are in
point_cloud_processing
To install ROS you can reffer to [ROS](https://www.ros.org/)
- To run the data processing scripts you need the following dependencies in ROS.
#CV Bridge to link the OpenCV to ROS
sudo apt install ros-kinetic-cv-bridge
- Other local dependencies are:
#OpenCV
pip3 install opencv-python
#Pypcl
pip install PyPCL==0.1.9
- The drivers for the sensors used
# HD webcam (Logitech webcam carl zeiss tessar)
(https://github.com/gautamjain1009/R-D-dataCollection) under folder named `my_camera`.
# Hypersen 3d solid state Lidar
(https://github.com/ropod-project/hps_camera)
# Intel Realsense d435 3d camera
- can be found under (https://github.com/IntelRealSense/realsense-ros) follow the instructions written there to install the driver.
-
Download the KITTI Depth Dataset from their website. Use the following scripts to extract corresponding RGB images from the raw dataset.
-
Pre-processed data collected with Intel Realsense. (for raw data contact the author) (https://drive.google.com/drive/folders/1ocrQCeEgyUfY9WrD8xkUOHOy6bDucdoX?usp=sharing)
-
Pre-processed data collected with Hypersen 3d Solid State lidar and RGB camera. (for raw data contact the author) (https://drive.google.com/drive/folders/1jhHfoTIcKJP5HG3gOxz3KfdxaOirw4rq?usp=sharing)
- To execute the data processing commands for collected data
#Download the bag files collected on Intel Realsense or Hypersen and camera setup.
rosbag play 20191122_1.bag #replace the name of the bag file
# To run the rosnode for data processing for data collected on Hypersen and camera setup.
rosrun point_cloud_processing dataScript.py
# To run the rosnode for data processing for data collected on Hypersen and camera setup.
rosrun point_cloud_processing data_script_realsense.py
- Data resizing and RGb and depth allign scripts are
R-D-dataCollection/point_cloud_processing/src/resized_allign_dataset.p,R-D-dataCollection/point_cloud_processing/src/allign_rgb_depth.py
# To run the RGB and depth allign script
python resized_allign_dataset.py
I can not create a comman script or an environemtn for all the approaches together as the dependencies are different and comman dependencies are in different versions.
- The specifications files can be found in (https://github.com/gautamjain1009/R-D-dataCollection) under
R-D-dataCollection/Environment_specifiaction_files.