Hacker News Reader

I did work on optical sorting machine: you have stream of fruits on very fast conveyor belt and then machine vision system scans the passing fruits, detects each object and reject (by firing stream of air) those that don't pass: those can be molded fruits, weird color or foreign material like rocks or leaves. 100 fps might be enough, but faster you go, faster your conveyor belt could be.

The 100fps model is also much more efficient in W/Flop, or J/Flop, or W/fps which is very important for embeded and mobile applications. You can design your construction site surveillance system to record 10 frames a minute while sleeping the ML accelerator and then process 100 frames all at once in a few seconds, which reduces the duty cycle tremendously, improving battery / device life.

Barcode scanners routinely perform 100 scans per second.

Of course, some would say machine learning is needless complexity if you just want to scan barcodes :)