Items with no label
3335 Discussions

Intel Realsense R200

cprak1
Beginner
3,023 Views

Hi!!!

I am very new to Intel Realsense Camera. My objective is to create an application to detect an object by capturing the image of the object and the background separately and subtract them to get only the object.

I got to know that i can make this using opencv and Librealsense library in visual studio. But I am not sure wheather it is the right path towards my Obective. Someone please provide some insights to me.

If this is right , next comes the problem for me....

I have already installed opencv and downloaded Librealsense from

https://github.com/IntelRealSense/librealsense.git GitHub - IntelRealSense/librealsense: Cross-platform API for Intel® RealSense™ devices

But I have no idea on how to install this Librealsense in my computer. Also i couldn't find any helpful information on installing this in Windows.

Any help would be really appreciated...

0 Kudos
16 Replies
MartyG
Honored Contributor III
688 Views

With the R200, you can get an object using a method called Object Recognition. If you wish to use Windows, I believe that it would be better for you to access this method with version '2016 R2' of the official RealSense SDK instead of using Librealsense.

When the RealSense R2 SDK is installed, it places a folder on your desktop called 'Intel RealSense SDK Gold'. This folder contains a 'Sample Browser' application, with which you can browse sample programs.

In the 'R200' section of the Sample Browser is a sample program called 'Object Recognition'.

You can download the '2016 R2' RealSense SDK with the link below, which launches automatically in your browser as a 1.8 GB download when clicked.

http://registrationcenter-download.intel.com/akdlm/irc_nas/vcp/9078/intel_rs_sdk_offline_package_10.0.26.0396.exe http://registrationcenter-download.intel.com/akdlm/irc_nas/vcp/9078/intel_rs_sdk_offline_package_10.0.26.0396.exe

cprak1
Beginner
688 Views

Thank you such for your response... I will try this one and update you about my progress!!!

0 Kudos
cprak1
Beginner
688 Views

Hi MartyG,

I have installed the SDK as you mentioned.

Now I can able to use this sample applications. i will explain my objective clearly..

First i need to capture an image without an object (Here object means a product from a company which manufactures 100s of different products . From that i need to identify what kind of product it is ) and then i have to capture an image with an object and i have to subtract both the images in order to get only the object and i have to display it to the user. Also i have to show the depth information of how far the object is from the camera.

So How can i proceed ? I have average c++ knowledge. What are the other things should i learn?

Please help me by giving some clues from where should i start?

I would be very much grateful if you can help me.....

0 Kudos
MartyG
Honored Contributor III
688 Views

The R2 SDK can recognize a list of 33 kinds of objects, which are detailed on this page.

https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?doc_or_object_recognition.html Intel® RealSense™ SDK 2016 R2 Documentation

So if the products in that company's range fall into the categories of objects listed on this page then RealSense may be able to recognize them. I would assume that it does this by recognizing characteristics of an object that looks like a chair, laptop, etc. Otherwise it would not be a useful feature if it could only recognize objects that looked exactly like ones that the developer of the Intel object recognition feature had programmed.

If the company's products do not fall within this list of 33 object types then that will be a problem. As support for the R200 has now ended in the Intel RealSense SDK, further objects will not now be added to this list. So you would need a way to train the SDK to identify objects that you have defined yourself. This would mean switching to the Linux version of the RealSense SDK, which has a tool called YOR that lets you define your own objects.

https://software.intel.com/en-us/documentation/realsense-for-linux-developers-guide/yor-tool Your Object Recognition (YOR) Tool | Intel® Software

Regarding the depth sensing ... I am guessing that you mean that you want to get the distance of a physical real-world object, and not a picture of an object like one in a catalog. You can get the real-world distance coordinates using the ProjectCameraToDepth instruction.

https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?projectcameratodepth_pxcprojection.html Intel® RealSense™ SDK 2016 R2 Documentation

The background could be removed using the Background Segmentation feature. There is a C++ sample in the R200 section of the R2 Sample Browser called SegmentationEP.

0 Kudos
cprak1
Beginner
688 Views

Thank you somuch. It will be very much useful for me. Realsense SDK has many samples in it , I cannot use the whole SDK for my application . So how can i make a seperate application only for performing my Objective. Can you please elaborate the steps to create our own application?

0 Kudos
MartyG
Honored Contributor III
688 Views

The first step is to set up your RealSense C++ development environment in Visual Studio C++

https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?doc_devguide_cpp_application.html Intel® RealSense™ SDK 2016 R2 Documentation

0 Kudos
cprak1
Beginner
688 Views
0 Kudos
MartyG
Honored Contributor III
688 Views

You are welcome! Please come back to the forum any time that you need further assistance. Good luck!

0 Kudos
cprak1
Beginner
688 Views

Hi MartyG,

First of all thank you so much for keep on helping me......... Here comes some new problem

The idea of using https://software.intel.com/en-us/documentation/realsense-for-linux-developers-guide/yor-tool Your Object Recognition (YOR) Tool | Intel® Software this software wont be good if the number of objects are more than 10000 and also if it is changing from time to time (ie.. there might be the possiblity of new objects coming everyday and we cannot able to differentiate whether it is a new object or the old object which we have already programmed using this tool ) So that is the reason that i could not able to convince my boss .. So i have to think in some other way !!!!

So as of now, i need to know how can we capture an 3d image using R200 camera and save that in our Hard disk ?

Also.... is there any sample programs or some stuffs with which i can program this and that image should be made available to the user , who can use it for later processing...

I would be very much grateful if you can help me.....

0 Kudos
MartyG
Honored Contributor III
688 Views

The R200-compatible sample program RF_ScenePerception (C++) supports saving a scan as an OBJ format model file.

In the 3DScan (C# ) sample, you can also save a scan as an OBJ after clicking the 'End Scanning' button.

Once you have an OBJ file, you can also import it into a 3D modeling package such as Blender for further editing.

0 Kudos
cprak1
Beginner
688 Views

Thank you so much!!!

i will try this and get back to you!!!

0 Kudos
cprak1
Beginner
688 Views

Hi MartyG!!

Now i have switched to SR300 camera for better accuracy. i need to take a picture(not a video) with depth information and i have to save the picture in Bitmap format and further which tool can i use to acquire the details from the image.

How can i proceed these with SR300 camera?

Expecting your valuable feedback..

0 Kudos
MartyG
Honored Contributor III
688 Views

It is possible to save as a bitmap but it is not an easy task, unfortunately.

The link below is the best information I know of for taking an image from the official RealSense SDK in Windows and converting it into a bitmap. The person asking the original question in the discussion said that they were taking the image from RealSense's "Camera Viewer" program.

https://stackoverflow.com/questions/32174076/how-to-save-an-image-in-intel-realsensevisual-c How to save an image in Intel RealSense(Visual C++) - Stack Overflow

If they are using the Windows SDK though then I think they probably meant the Camera Explorer app.

https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?doc_tools_tool_camera_explorer.html Intel® RealSense™ SDK 2016 R2 Documentation

0 Kudos
cprak1
Beginner
688 Views

Thanks a lot for your valuable feedback... i will try it and get back to you...You have said that the task of saving the image in Bitmap is quite difficult. So can you suggest me the other possiblitiy of saving the image file to use it further without compensating the quality and details of the image?

0 Kudos
MartyG
Honored Contributor III
688 Views

The other approach I have suggested to people is to do the conversion manually, by recording the camera data as an .RSSDK format video file (which has the .h264 type of video encoding) and then converting the .h264 video file into a bitmap.

0 Kudos
cprak1
Beginner
688 Views

Thank you so much for giving qucik and prompt answer.... I will try out these things and will come back to you...

0 Kudos
Reply