Amarjot Singh

Introduces a real-time drone surveillance system that uses Feature Pyramid Network to detect humans from aerial images and identifies violent individuals in public areas.

Addresses the challenging problem of disguised face identification using a deep learning framework that detects facial key-points to improve identification accuracy.

Introduces a novel ScatterNet that uses a parametric log transformation with Dual-Tree complex wavelets to extract translation invariant representations for object classification.

Proposes a DTCWT ScatterNet Convolutional Neural Network (DTSCNN) that replaces early CNN layers with a parametric log based DTCWT ScatterNet to extract edge-based invariant representations, improving network training.

Introduces the ScatterNet Hybrid Deep Learning (SHDL) network which extracts invariant and discriminative image representations for object recognition, featuring a multi-layer ScatterNet front-end, an unsupervised learning middle, and a supervised learning back-end module.

Presents a method for accurate assessment of embryo viability by automatically segmenting the trophectoderm in microscopic images of human blastocysts.

Proposes an efficient wavelet-based algorithm for brain tumor detection that utilizes complementary and redundant information from CT and MRI images for accurate size and location detection.

A comprehensive review of Optical Character Recognition (OCR) applications, highlighting its wide use in various fields.