A better methodology would be to do a physical audit. Have some people in each city manually count the number of cameras over some area, then extrapolate from there. This has the flaw of missing hidden or inaccessible cameras, but at least would be consistent.
Why? Multiple sources shouldn't matter, it is the quality of sources. If you're gathering data you shouldn't expect it to all come from one place. There's no single source that tracks every single thing. At some point you're going to have to gather data from multiple sources, especially if you're trying to make comparisons that others haven't.
This assumes a uniform distribution where one isn't obvious. Some places are bound to have more security than others.
I don't think it matters much though, as a simple count of cameras isn't so useful (especially with them being so cheap now). How many cameras is less important than what they cover and what they are connected to.
Solving the long tail end has decreasing returns.
If you already have a well-behaving society, then adding a few cameras probably won't change anything at all. You need a whole lot of cameras to have any meaningful effect.
Is it part of the zero-unemployment busywork culture?
3. Shanghai, China – 2,985,984 cameras for 26,317,104 people = 113.46 cameras per 1,000 people
76. Tokyo, Japan - 24,500 cameras for 37,435,191 people = 0.65 cameras per 1,000 people
Well, first off what are we comparing? They list 26.3 million people for Shanghai and 37.4 million for Tokyo. That's comparing the metropolitian area of Tokyo (Tokyo, Kanagawa, Saitama, Chiba) to just the city of Shangahi. Shanghai's metropolitian area population is closer to 33 million.
If we change Tokyo to the city population (9.3m) which is what they used for Shanghai then Tokyo jumps to position 40 at 2.66 cameras per 1000 people.
On top of that 24,500 seems awefully low for Tokyo. There are cameras everywhere and have been since I first got here 21 years ago. I always found it strange if the Japanese are so honest (perception may be different than reality) then why do they need so many cameras? Of course I haven't counted, just saying for a city so large and seeing the cameras stick out if you're looking it feels low.
All of that is the long way of saying the data is highly suspect.