- Q2.ipynb
- Implemented
2D convolutionand2D max poolingfunctions using only numpy. - Invoked functions on a simple image using the given kernels.
- Implemented
- Q3.ipynb
- Transformed images into frequency domain using
2D DFTand2D FFT. - Implemented low-pass and high-pass filters and Applied them to the images.
- Transformed images into frequency domain using
def Conv2D(img, filters, stride=1, padding='same'):
.
.
.
return output_image| Parameters | Type | Detail |
|---|---|---|
| img | np.ndarray |
height * width * channels |
| filters | np.ndarray |
f * f * number_of_filters |
| stride | Integer |
1 by deafult |
| padding | String |
"same" or "valid" |
def MaxPooling2D(img, window=(2, 2), stride=2):
.
.
.
return output_image| Parameters | Type | Detail |
|---|---|---|
| img | np.ndarray |
height * width * channels |
| window | Tuple |
(height, width) |
| stride | Integer |
2 by deafult |
