Technology
This is the official technology community of Lemmy.ml for all news related to creation and use of technology, and to facilitate civil, meaningful discussion around it.
Ask in DM before posting product reviews or ads. All such posts otherwise are subject to removal.
Rules:
1: All Lemmy rules apply
2: Do not post low effort posts
3: NEVER post naziped*gore stuff
4: Always post article URLs or their archived version URLs as sources, NOT screenshots. Help the blind users.
5: personal rants of Big Tech CEOs like Elon Musk are unwelcome (does not include posts about their companies affecting wide range of people)
6: no advertisement posts unless verified as legitimate and non-exploitative/non-consumerist
7: crypto related posts, unless essential, are disallowed
view the rest of the comments
But like... why? This seems like it solves a problem we don't have - hybrid CV systems are already far more efficent than what they're describing here. Feeding the token input directly into the transformer is... what people try to avoid doing, as that uses far more energy than you'd ever consume just transfering images from a CCD. The tokenizer is not going to be the source of significant inefficiency in an AI-enabled system...
๐คก
A cutting wit unique in its originality, truly. I may never recover. You're welcome to address the question tho - this tech seems like a novelty more than anything.
Ah yes, another armchair expert who clearly knows more than the researchers at Nanjing University and the editors of Nature Sensors.
Power constraints are not a concern. Drones and satellites just have unlimited batteries, right? They definitely don't need to shave off every milliwatt possible.
Hybrid CV systems are already perfect. Why would anyone bother innovating when we can just keep moving massive amounts of redundant data through multiple conversion stages? It's not like that burns energy or anything.
But sure, your armchair analysis definitely outweighs peer reviewed research from an international team that built a working prototype with molybdenum disulfide phototransistors.
They probably just forgot to Google 'hybrid CV' before publishing.
.... You feeling okay? It's really dumb to be crashing out this hard over popsci reporting. Maybe go give the actual paper a read? It doesn't answer my criticisms, hence presenting them here, but at least we'll have a common basis on which to discuss my criticism of the research.
(well... they did offer me a job, so I suppose they must see something of value in my work? It was pretty far outside my field though. And, honestly, they're just shotgunning out offers to any researchers that publish)