CAM Mask Shape

[oseymour]

Hello! I'm encountering the following error:

ValueError
This cell raised an exception: ValueError('operands could not be broadcast together with shapes (982,748,3) (748,982,3) ')

Here's my code:

<relevant imports here>

model = YOLO(<path to torch file>)

target_layers = [model.model.model[-2]]

cam = EigenCAM(model, target_layers, task="cls")
grayscale_cam = cam(
    frame,
    eigen_smooth=True,
)[0, :, :]

cam_image = show_cam_on_image(frame / 255, grayscale_cam, use_rgb=True)
plt.figure()
plt.imshow(cam_image)
plt.show()

My input, frame, has the correct dimensions of (784, 982, 3). When I view the CAM mask on it's own it has the right orientation but the dimensions are not correct. When I try transposing it, the dimensions match but then the orientation is not correct.

For now I'll just resize the CAM mask but I thought you might want to know the output dimensions are not correct.

Let me know if you want my input files or any other info to debug this. It's a project at work so I can't share all details but I can try to find public data that replicates my issue.

Thank you! Your repo is meeting my exact needs!