Yolo_eval problem

Incompatible shapes: [19,19,5,4] vs. [19,19,5,80] [Op:Mul]

Happens inside yolo_eval

[code removed - moderator]

Please click my name and message your notebook as an attachment.

Here’s how I would start to reason about this problem.

What object has a shape of (19,19,5,4)?
What do each of those dimensions (19x19, 5, and 4) represent?

Same for the object with shape (19,19,5,80).

Spoilers:
19x19 is the number of grid cells
5 is the number of bounding boxes
4 is the number of bounding box predictions (two for the center location, two for the shape)
80 is the number of classes

So in the multiplication operation throwing the exception, the bounding box coordinates are being multiplied by the class predictions. Does that make sense? What values available at that step make more sense? Any hints provided in the notebook cells describing the problem?