Gurugram, Haryana, India
Improved pre-existing facial recognition model by reducing the size to 26% and ported it to embedded hardware which ran it at 8 FPS. Ported pre-existing human classification pipeline to tensorflow.js resulting in a 70% cost reduction.