Conv Patch Embedding

Problem

Apply a patch embedding as an unfold plus dot product.

For every non-overlapping patch (row-major) compute the sum of pixel * weight over the flattened patch, where weight has patch * patch entries. Return the list of outputs rounded to 4 decimal places.

Test cases

case 1

in: [[[1,2],[3,4]],2,[1,1,1,1]]

expected: [10]

case 2

in: [[[0,1,2,3],[4,5,6,7],[8,9,10,11],[12,13,14,15]],2,[1,0,0,0]]

expected: [0,2,8,10]

case 3

in: [[[1,2],[3,4]],2,[1,2,3,4]]

expected: [30]

case 4

in: [[[1,2,3],[4,5,6],[7,8,9]],3,[1,1,1,1,1,1,1,1,1]]

expected: [45]

Starter codePython

def conv_patch_embed(image, patch, weight):
    # Your code here
    pass

4 test cases · run your code against them in the editor