Nir Regev

Ph.D. EE, research scientist, with >25 years of experience in developing algorithms, with proven ability to develop practical solutions to "problems that can't be solved". Specializing in radar signal processing, computer vision and deep learning. Companies I worked for include Essence, Intel, Xilinx, Rada, and IDF. __________________________________________________________________________________ Some of the algorithms and projects I implemented are : Radar signal processing, tracking, estimation and Detection : CFAR detection, adaptive beamformers, DOA estimation (Maximum Likelihood, MUSIC, root-MUSIC), Kalman filtering, source enumeration (BIC, MDL, AIC, etc.), deep nets for radar target classification. __________________________________________________________________________________ Machine vision: Deep learning models for target detection and identification, pose estimation, foreground/background segmentation, clothes segmentation, 3D reconstruction, stereo vision, automatic and blind image distortion calibration algorithm based on Brown’s distortion model and iterative nonlinear optimization (Gauss-Newton’s algorithm). Facial recognition, object recognition and classification, vital signs extraction from video. __________________________________________________________________________________ Machine Learning and deep learning: MLP, CNN, RNN (LSTM), Support Vector Machine (SVM), Bayesian classifiers (GLRT), kNN, Multi-Edit and Condensing algorithms for kNN classifiers, Gaussian Mixture Estimation, LDA, PCA and other Model order Selection Algorithms like BIC, MDL, AIC, GIC and more.. __________________________________________________________________________________ Remote sensing: Remote vital signs extraction using a low-power, mmWave radar, human gait analysis and classification using low cost ultrasonic transducers