How AI Helped Improve Crowd Counting in Hong Kong Protests (nytimes.com) 43
K.K. Rebecca Lai, Jin Wu and Lingdong Huang, writing for the Times: Crowd estimates for Hong Kong's large pro-democracy protests have been a point of contention for years. The organizers and the police often release vastly divergent estimates. This year's annual pro-democracy protest on Monday, July 1, was no different. Organizers announced 550,000 people attended; the police said 190,000 people were there at the peak. But for the first time in the march's history, a group of researchers combined artificial intelligence and manual counting techniques to estimate the size of the crowd [Editor's note: the link may be paywalled], concluding that 265,000 people marched. The high density of the crowd and the moving nature of these protests make estimating the turnout very challenging. For more than a decade, groups have stationed teams along the route and manually counted the rate of people passing through to derive the total number of participants. Though the use of A.I. does not make the calculation definitive, the technology helps produce a more precise estimate because it uses computers to try to count every person.
Since 2003, Paul Yip, a social sciences professor at Hong Kong University, has been producing a count of the size of protests held annually on July 1, the anniversary of Hong Kong's 1997 handover from Britain to China. With the hopes of creating a more robust estimate this year, Mr. Yip teamed up with Edwin Chow from Texas State University and Raymond Wong from C&R Wise AI, a local technology company, to use artificial intelligence to count the crowd at the march. Using open source software, The New York Times developed a computer model to illustrate how artificial intelligence could be used to recognize people and objects moving within a video. Analyzing a short video clip recorded on Monday, The Times's model tried to detect people based on color and shape, and then tracked the figures as they moved across the screen. This method helps avoid double counting because the crowd generally flowed in one direction.
Since 2003, Paul Yip, a social sciences professor at Hong Kong University, has been producing a count of the size of protests held annually on July 1, the anniversary of Hong Kong's 1997 handover from Britain to China. With the hopes of creating a more robust estimate this year, Mr. Yip teamed up with Edwin Chow from Texas State University and Raymond Wong from C&R Wise AI, a local technology company, to use artificial intelligence to count the crowd at the march. Using open source software, The New York Times developed a computer model to illustrate how artificial intelligence could be used to recognize people and objects moving within a video. Analyzing a short video clip recorded on Monday, The Times's model tried to detect people based on color and shape, and then tracked the figures as they moved across the screen. This method helps avoid double counting because the crowd generally flowed in one direction.
How? (Score:3)
It just identified every single one of them from their face, their gait, their jeans* and ears**.
(the last 3 for the masked ones)
Then it did a sum(above).
*Wired had an article about it 20 years ago.
**8 years ago
Re: (Score:2)
Hong Kong was leased by the Brits from China for 100 years after the former won the Opium War. When Britain negotiated Hong Kong's return in 1984, there were a number of factors that made it a fait accompli. For one, having stretched themselves just fighting Argentina over the Falklands, it was clear that the UK couldn't possibly prevent Beijing from doing something similar if they didn't come to an agreement. So they had a negotiation w/ China over Hong Kong, and managed to save face by agreeing to a 'O
Re: (Score:2)
Thanks for correcting me about what was ceded and what was leased. I was under the impression that all of it was the latter. Also, thanks for letting me know that it was the Ching empire.
I understand the misgivings about Hong Kong being handed over, but like I pointed out in the GP, if we look at it from the British POV, they didn't really have good options. A year earlier, Argentina had militarily seized the Falkland Islands, it took them some days for the first ships of their navy to get there and a
Re: (Score:2)
Re: (Score:2)
"if we look at it from the British POV, they didn't really have good options. "
They never had those, not in Palestine, not in India/Pakistan, Iran/Irak and now with Brexit.
We can be sure that they will fuck that up as well.
Re: (Score:2)
Re: (Score:2)
"If the current display of support for democracy by the Hong Kongers is genuine, then they should demand independence from China"
Ah, like Catalunia got their independence from Spain, just by asking for it.
Assisting (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Came here to say exactly this!
Re: (Score:2)
Re: (Score:2)
It's the 5th. Of course it's a slow work day.
Sorry, I couldn't help myself.
Re: (Score:3)
More like determining who to toss into an actual concen...sorry I mean 're-education' camp.
Re: (Score:2)
I'm assuming it is also assisting with the arrests.
So far, only a few thousand are known to be missing.
The Chinese authority purposely allow the riot (Score:1)
It's a fishing exercise.
If the Chinese authority does not allow any riot they wouldn't know who will take part.
Hong Kong is wired, just like the rest of China's cities, and the authority is watching, and recording every movement of everybody every second of the time.
Now the Chinese a authority has in their hand 1. The number of rioters, 2. The identities of the rioters, 3. The leaders of the rioters, 4. Where and how they gather.
With big data they can carry out a further trace on t hose rioters, and with AI
it's time for big data to cut China off! trump jus (Score:2)
it's time for big data to cut China off! trump just needs to say that big data ceo's can face the death penalty for any one that china executions.
Just the idea alone can be used in an trade deal
Re: (Score:2)
This is no different than what is done for pipeline protesters in America, or any type of protest that doesn't support the big picture of our
government.
Re: (Score:2)
Now the Chinese a authority has in their hand 1. The number of rioters, 2. The identities of the rioters, 3. The leaders of the rioters, 4. Where and how they gather.
With big data they can carry out a further trace on t hose rioters, and with AI they can locate the so-called 'missing links' amongst the different groups of rioters.
The news report is that Hong Kong authority has picked up 28 rioters.
One thing about these protests - unlike other protests in the past, these ones have no leaders. Which is fortunate, since there ain't a 'choke point' by which to crack down on the entire movement. So no matter who gets killed, that one would continue, until everybody involved has either fled or gotten killed/captured
Re: (Score:2)
Simple software we call an app.
Complex software we call AI.
Could probably just call it a linear algebra algorithm.
Counting watermelons (Score:2)
Last year, there was an election contest in one of the counties in Taiwan that was highly contested. It drew so much interest that many of the rallies that opposition parties started to play down the attendants. However, several industrious computer science students implemented an algorithm that was used for counting crop yields (lettuce and watermelon) and retrained it to count heads of people. Drone laws in Taiwan is pretty relaxed so flying drones over people right now is pretty much an everyday thing es