stanfordvl / cs131_release Goto Github PK

• the autograder tests often report false is not true - would be helpful to have more meaningful feedback in autograder tests (ie. providing the computed output compared to the reference output)

I've been doing these exercises on my own but couldn't get the test code with the 3x3 matrix in the forward energy part of the problem set to pass. Upon closer inspection, it looks like the forward energy may be computed with the difference in energy, not the difference in image intensity as the lecture notes say it should be. Am I wrong in my understanding of how this is supposed to work? For instance I expect the solution cost for the middle value of the 3x3 test matrix to be the minimum of 2.5+0.5=3, 0.5+0.5+0.5=1.5, 3+0.5+1=4.5 for vertical, left and right seams, which is 1.5 yet the solution cost claims it to be 2. The only way I can get 2 for this cost is if I compute the new introduced energy as the difference in energy, not the difference in pixel intensity. I can make a PR for it if I'm not mistaken here, although I also don't really know what the boundary conditions are supposed to be in this case.

• Many students having issues in RANSAC that don’t appear until the stitch multiple images step, add a clearer sanity check on RANSAC, (ie. Printing out the keypoint values or H values), or give students some form of better intuition for debugging the question.

• Do the same on split_multiple_images (sanity check the H values and the composite matrix shape before having them move on to the next step(s))

• 1.2 (b) is unclear after removing the proof above from last year (show that $(K^{-1}v_i)^\top(K^{-1}v_j) = 0, \text{for each } i \neq j.$). I'd recommend putting this after 1.3 (c) or removing the question completely

Quick question on the HW2 problems from Section 1.3 on Non-maximum suppression.

Everything so far makes sense except for the second step of the algorithm:

2. Compare the edge strength of the current pixel with the edge strength of the pixel in the positive and negative gradient directions. For example, if the gradient direction is south (theta=90), compare with the pixels to the north and south.

What I'm not understanding is when theta=90, why is the gradient direction south and not north? It seems as though theta is being measured clockwise from the positive x-axis. Why?

Hello, I am a beginner. In hw3_release Extra Credit: Stitching Multiple Images, I don't know how to do.
The problem said given a sequence of m images, then stitch an ordered chain of images. I want to take the first and second images then stitch together, formed a new images, then stitch together the third images. But it is too complex and maybe have some simple ways.
I notice that you haven't do this problem, if you have any idea, please complete the code. I am a beginner, it is too hard for me. Thank you so much.

The above slide comes from lecture-06.
I assume the eigenvectors determine the change direction, but what lamda^(-0.5) mean while lamda is eigenvalues?
any sugguestion is welcomed
Thanks

How to load references?
I dont see any file release

either google cant find it...
thanks