Neural network delivers re-identification in video surveillance with only 2.2 million parameters
Researchers at the University of Surrey have developed a lightweight neural network that only needs 2.2 million parameters for re-identification of people in video