- Implementation of NCC in numpy
- aligning images based on correlation
- implementation of scale-normalized Laplacian of Gaussian (LoG) operator
- building LoG pyramid
- finding blobs from local maxes
- detecting keypoints using SIFT
- estimation of putative matches between local descriptors of 2 images
- usage of RANSAC (because of false-matches) to estimate transformation between images
- join image based on transformation matrices
- extract features from images
- create clusters of image features (use KMeans to create bag-of-words model)
- use k-NN to predict class based on manually extracted features
- Implementation of PyTorch model to classify images
- Dataset (augmentation, loading)
- Trainer (losses, optimizers, training in general)
- Classifier (NN architecture)