Fuzzy visual encoding using Gustafson-Kessel and Gath-Geva clustering. Learns adaptive covariance matrices per cluster for soft SIFT descriptor assignment. Outperforms k-means BOF on Caltech and VOC benchmarks. Published at ICIP 2012.

This project utilizes logistic regression to classify numbers 0 and 1 using sign language gestures. It successfully achieves the task of sign language classification, reaching a test accuracy of 93.54%.

Subspace projection methods (PCA, SPCA, SSPCA, KernelPCA, Isomap) for visual classification. Structured Sparse PCA with co-clustering outperforms baselines on VOC and Caltech benchmarks. Published at BigMM 2017.