Computer Vision News - August 2019

Spotlight News ‰ ‰ ‚ Ž The Google AI blog does not publish many articles, but they are all very good. This one tells us that to improve models, a better understanding of generalization is required, which can lead to more theoretically grounded and therefore more principled approaches to DNN design. “ “ “ “  ‚ Ž Ever wondered how do throwing athletes get their tools back after the throw? I can’t tell about previous Olympics, but I know about Tokyo 2020 games: Toyota is developing a little self-driving robot with enough AI to fetch javelins and hammers. It seems that humans are better suited to picking up heavy equipment, but robots are a better choice for quickly transporting them to their respective owners.  ” ˆ ˆ • We believe there is a third way and Forbes does too: besides buying a packaged AI or developing it from scratch, you can outsource and get the best possible AI for your needs. In this way, you gain access to an already hired, trained and managed team of technology experts. You save headcount, you benefit from all their past experience and the AI is yours. ˆ – —  — ƒ “ “ “ “    “““ 37