FilterBank_Python

python code for generating Filter Bank using for image retrieval. Contents including:

Derive 2D Gaussion filter and Laplacian of Gaussion (LoG) (in report.pdf)
Generate a filter bank of 48 image filters, each with 32 & times 32, including 4 Gaussion filters, 8 LoG filters, 18 x-directional first derivation of Gaussian filters and 18 x-directional second derivation of Gaussian filters
Retrieve top5 images given a image query using filter bank and deep features

Filter Bank (48)

Filter Bank features

Deep Features from ResNet101

Name		Name	Last commit message	Last commit date
Latest commit History 4 Commits
Assignment1_data/Assignment1_twoImg		Assignment1_data/Assignment1_twoImg
Images		Images
Convolution.ipynb		Convolution.ipynb
FilterBank.npy		FilterBank.npy
ImageRanking.ipynb		ImageRanking.ipynb
QueryRes101Features.npy		QueryRes101Features.npy
QueryRes50Features.npy		QueryRes50Features.npy
README.md		README.md
Res101Features.npy		Res101Features.npy
Res50Features.npy		Res50Features.npy
features.npy		features.npy
filePath.npy		filePath.npy
rankList.txt		rankList.txt
report.pdf		report.pdf
resnet_test.py		resnet_test.py