Here is a collection of images of various blocks world structures.
Download OpenCV from here. Installing for python Installing for python 2
I put together some OpenCV examples. Unzipped directory
Contents:
Makefile - instructions on how to compile these programs on Linux CGA programs: pyr-blobs-color.cpp - current program - build pyramid - find variances - find blobs (areas of low variance) - use LSD line detector - use Canny line detector pyr-blobs.cpp - early attempt at blob finding pyr-edges.cpp - early attempt at edge finding find-colors.cpp - find which colors are used most Opencv demo programs: pyramid.cpp - Image pyramids https://docs.opencv.org/4.0.1/d4/d1f/tutorial_pyramids.html edges.cpp - Canny edges https://docs.opencv.org/4.0.1/da/d5c/tutorial_canny_detector.html hough.cpp - Hough transform hough2.cpp https://docs.opencv.org/4.0.1/d9/db0/tutorial_hough_lines.html lsd-lines.cpp - Line segment detector http://visilab.etsii.uclm.es/personas/oscar/Publications/New_functionality_OpenCV3.pdf bounding.cpp - Bounding boxes https://docs.opencv.org/4.0.1/da/d0c/tutorial_bounding_rects_circles.html contours.cpp - Find contours http://opencvexamples.blogspot.com/2013/09/find-contour.html https://docs.opencv.org/4.0.1/d4/d73/tutorial_py_contours_begin.html harris.cpp - Find corners https://docs.opencv.org/4.0.1/d4/d7d/tutorial_harris_detector.html shi-tomasi.cpp - Find corners https://docs.opencv.org/4.0.1/d8/dd8/tutorial_good_features_to_track.html segment1.cpp - Segmentation using the watershed approach https://docs.opencv.org/4.0.1/d3/db4/tutorial_py_watershed.html
BLORT -The Blocks World Robotic Vision Toolbox. See this version, which may be more up-to-date.
Localizing 3D Cuboids in Single-view Images