Open nghorbani opened 1 month ago
wierd is that they sometimes (at certain input values) come very close
the 3D filter looks meaningful
input_arr = np.random.rand(10,1,3) # (100,1,3) print(f"Input array: {input_arr.shape}") output_arr = filter3D(input_arr, mu=7, poly=5, order=0) # (100,1,3) plot_3d(input_arr, output_arr)
Hi, I wrote a simple test and the output size is different ` input_arr = np.random.rand(10,1) output_arr = filter1D(input_arr.reshape(-1), mu=5, poly=3, order=0)
` resulting in output_arr of shape 10, and output_arr2 shape of 5. it has to do with conv(...,'full'). if "same" is used the dimension are correct but still the values are different