Lab 4

[tutoringjedi]

What changes are you trying to make? (e.g. Adding or removing code, refactoring existing code, adding reports)

Completing Lab 4 (Convolutions): exploring how Conv2D kernel size and padding affect output shape, implementing an edge-detection kernel (Laplacian) on a grayscale image, applying max/average pooling, and classifying images (a sample cat photo and a webcam snapshot) using a pre-trained ResNet50 model.

What did you learn from the changes you have made?

padding="valid" shrinks the output's spatial dimensions by kernel_size - 1 in each direction, while padding="same" zero-pads the input so the output stays the same size regardless of kernel size. A larger kernel also means each output pixel mixes a wider neighborhood of input pixels, giving a more smoothed-out result. A Laplacian kernel highlights edges in every direction at once (unlike Sobel, which needs separate horizontal/vertical kernels), by responding strongly to local intensity changes and staying near zero over flat, uniform regions.

Was there another approach you were thinking about making? If so, what approach(es) were you thinking of?

Considered using a Sobel kernel pair (separate horizontal and vertical kernels, combined via magnitude) for edge detection instead of a single Laplacian kernel, which would give directional edge information rather than an omnidirectional response.

Were there any challenges? If so, what issue(s) did you face? How did you overcome it?

Hit a ModuleNotFoundError for cv2 when trying to use the webcam capture section; resolved by installing opencv-python (the pip package name doesn't match the cv2 import name, which tripped me up initially).

How were these changes tested?

A reference to a related issue in your repository (if applicable)

Checklist

• I can confirm that my changes are working as intended