Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Listening for illegal logging chainsaws using TensorFlow

blog.google

Post reply on HN
Print view View original

51–60 of 128 posts

Re: Listening for illegal logging chainsaws using TensorFlow

#51

Post by shaki-dora »

triangleman wrote: post #28
shaki-dora wrote: post #21

Earlier quoted context omitted.

This is a myth, mostly. First of all: no jail. This is civil law, not criminal. Second: it's pretty hard not to have an excuse to do anything you want if you're the board + management. Google, and most other companies, give to charitable organisations, for example. "Now", you'll say, "they're just doing this for PR". Maybe... But using the PR argument actually works perfectly fine for anything altruistic a company ma…

If a publicly owned company started to contribute too much money to charity, they would certainly be subject to shareholder lawsuits and a proxy war over management. Sick pay is easier to do because it's related to retaining good employees. But generally, as Milton Friedman famously wrote, "the social responsibility of corporations is to increase their profits".

"Apple donates $5 million to hurricane relief, makes it easier for customers to donate": http://money.cnn.com/2017/09/08/technology/apple-donation-hu...

(just search for [ donates], and you'll find examples of charitable donations for every single Fortune 500 company)

Re: Listening for illegal logging chainsaws using TensorFlow

#52

Post by bourhan-rfcx »

ada1981 wrote: post #45

I love these guys! I helped advise them on their crowdfunding campaign a few years ago. One of my favorite projects. Hey Topher! Are you guys cross checking this with daily satellite imagery? I met with a guy who's company gets a full HD shot of the earth every 24 hours and they'd be happy to give you guys access to the visual data (they also train algorithms on it to detect all sorts of activity). Anyhow, I'll shoot…

Thank you so much for you support!

We are not doing that at the moment but would love to hear more about it.

Re: Listening for illegal logging chainsaws using TensorFlow

#53

Post by pr0tonic »

ada1981 wrote: post #45

I love these guys! I helped advise them on their crowdfunding campaign a few years ago. One of my favorite projects. Hey Topher! Are you guys cross checking this with daily satellite imagery? I met with a guy who's company gets a full HD shot of the earth every 24 hours and they'd be happy to give you guys access to the visual data (they also train algorithms on it to detect all sorts of activity). Anyhow, I'll shoot…

What company is this if you dont mind saying?

Re: Listening for illegal logging chainsaws using TensorFlow

#54

Post by tony_cannistra »

stefanRfcx wrote: post #35
O_nlogn wrote: post #22

Earlier quoted context omitted.

This is a really cool application of this technology, great work! Do you have any blog posts explaining the types of models you're using or open source repos that one could contribute to? (Asking as a ML/Software Engineer who's pleasantly surprised and yearning to contribute to something like this!)

Thanks! We are a relatively small engineering team and we were mainly focused on improving our system. This took a lot of effort and time, we just haven't had a chance yet to think about open sourcing or blogging. Our days were focused on coding. If there's interest in blog posts, I could write up some of our technology. Just let me know whether there's an interest and what you'd like to know.

I know that the US National Park Service Natural Sounds/Night Skies division has an incredible about of environmental audio data that's begging for automated analysis. This could be of much use to them. They do analyse their data, but it's largely been manual [1].

[1]: http://science.sciencemag.org/content/356/6337/531

Re: Listening for illegal logging chainsaws using TensorFlow

#55

Post by tzahola »

loceng wrote: post #27

Are there methods to change the noise of a chainsaw, as there's money to be made in illegal logging - they'll have the incentive if they're organized, so it would be quiet enough or different enough to potentially sound like it's "outside" of a protected area or not a chainsaw at all?

Electric chainsaws!

Re: Listening for illegal logging chainsaws using TensorFlow

#56

Post by stefanRfcx »

tony_cannistra wrote: post #54
stefanRfcx wrote: post #35

Earlier quoted context omitted.

Thanks! We are a relatively small engineering team and we were mainly focused on improving our system. This took a lot of effort and time, we just haven't had a chance yet to think about open sourcing or blogging. Our days were focused on coding. If there's interest in blog posts, I could write up some of our technology. Just let me know whether there's an interest and what you'd like to know.

I know that the US National Park Service Natural Sounds/Night Skies division has an incredible about of environmental audio data that's begging for automated analysis. This could be of much use to them. They do analyse their data, but it's largely been manual [1]. [1]: http://science.sciencemag.org/content/356/6337/531

That's very interesting. The Rainforest Connection system has the ability to detect more than only chainsaws, so it could be really interesting to learn more from the datasets; we'll look into that. Thanks so much for sharing!

Re: Listening for illegal logging chainsaws using TensorFlow

#57

Post by tony_cannistra »

aynsof wrote: post #30

Someone more familiar with satellites can correct me, but from my understanding you should be able to detect deforestation from real-time satellite data. You alert on changes in the level of greenery in the images. The indicator might trail behind detecting chainsaw noises, but it wouldn't require the mass distribution of detector devices.

This is being done by many [ex. 1], but you've identified the challenge: often it's too late to catch offenders once the satellites have seen the damage.

[1]: http://data.globalforestwatch.org/

Re: Listening for illegal logging chainsaws using TensorFlow

#58

Post by zapfranklin »

pr0tonic wrote: post #53
ada1981 wrote: post #45

I love these guys! I helped advise them on their crowdfunding campaign a few years ago. One of my favorite projects. Hey Topher! Are you guys cross checking this with daily satellite imagery? I met with a guy who's company gets a full HD shot of the earth every 24 hours and they'd be happy to give you guys access to the visual data (they also train algorithms on it to detect all sorts of activity). Anyhow, I'll shoot…

What company is this if you dont mind saying?

I assume he is talking about Planet Labs (https://www.planet.com/) - a very cool service.

Re: Listening for illegal logging chainsaws using TensorFlow

#59

Post by bourhan-rfcx »

Rotdhizon wrote: post #23
RainforestCx wrote: post #17

Earlier quoted context omitted.

https://twitter.com/lilitjejauregui/status/97585519352188518...

That's for small time abuses. What happens in the case of a full blown illegal logging operation involving a entire company? What about cases where things get violent? It's not uncommon for mining operations in the Amazon to end with natives getting massacred, the loggers probably not unwillingly to do that to anyone who gets in there way. When you come across a full scale logging op, do things get complicated?

Great points you bring up and indeed things do get complicated as you face a bigger and more organized operation. This is the case in parts of the Tembe territory in Brazil and it is the reason why we work with many partners and agencies on the ground to ensure that the data we capture is used wisely.

Re: Listening for illegal logging chainsaws using TensorFlow

#60

Post by tonic_section »

stefanRfcx wrote: post #40
starchand wrote: post #11

Earlier quoted context omitted.

But can it identify lyrebird's?

That's a great question. Actually, one of the sounds that are pretty close to a chainsaw are mosquitos that are circling around our microphones due to the Doppler effect. We found ways of dealing with signals that are close to chainsaws by aggregating multiple models and also a time-based analysis. The system can draw causal/correlative conclusions such as a vehicle is usually present before a chainsaw. If there's no…

How do you quantify the confidence of your model? Do you use a Bayesian model or just the log-likelihood? Because the latter can act strangely in some cases.
Post reply on HN