Current solver has an error in line 3933 x10 = argmax(x9, x5). The code attempts to find the translate of maximum overlap with original object, but the function x5 simply computes the intersection, not the size of the intersection.
The second overlap computation in the solver is correct, using x14 = compose(size, x5) to find the argmax of size. The simplest fix is to just define x14 earlier, and use it in the first argmax instead of x5. My fix does this, and renames the variables sequentially.
Current solver has an error in line 3933
x10 = argmax(x9, x5)
. The code attempts to find the translate of maximum overlap with original object, but the functionx5
simply computes the intersection, not the size of the intersection.The second overlap computation in the solver is correct, using
x14 = compose(size, x5)
to find the argmax of size. The simplest fix is to just definex14
earlier, and use it in the first argmax instead ofx5
. My fix does this, and renames the variables sequentially.