What is quantization in deep learning, and why is it critical when deploying models to edge devices in a video telematics system like Lytx?
