Skip to content

cornell-cup/r2-edge-detection

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

86 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Testing Instructions

Install

To make sure you have all the required packages, you should create a virtual environment, activate it, and run

pip install -r requirements.txt (there might be more packages in there than necessary, and it might not have packages for files like kinematics.py, oops)

Important directories/files for testing:

Samples - contains testing images for 10 objects, each with 5 poses, from the JACQUARD dataset. The 10 subdirectories corresponds to the images for one object.

grasptest.py - contains the testing code that uses final.py's grab_points() grasp prediction function.

How to test:

Go to final.py and find where it says "Uncomment to see final image/edges/grasping points!". Uncomment the three lines below that.

python grasptest.py [dir_with_image_files] [take manual input True or False]

For each image, the code should display

  • What the RGB version of the image looked like after edge detection, along with grasp points (in pink)
  • What the original RGB image looked like, with grasp points

Upon closing the two display windows,

  • What the labels from JACQUARD were (orange for grasp points, yellow for grasp rectangle)
  • And the calculated grasp prediction. (The pink dots are the grasp points and the green rectangle represents the grasp rect with different sized end effectors)

After closing that window,

  • the image name and the mode [0 for "rgb", 1 for "d", 2 for "rgd"] will be printed, along with the automated validation result (which most of the time is not correct, sadly :( )
  • You will also be prompted to give manual input on whether the grasp prediction was valid.

After validating all the images in the specified directory, an array will be printed, with accuracy percentages corresponding to ["rgb", "d", "rgd"].

Use Ctrl + c (and close any open windows) to terminate running for any reason.

About

Edge detection algorithm for object gripping.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages