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

Chasing Cats

myplace.frontier.com

Post reply on HN
Print view View original

11–20 of 93 posts

Re: Chasing Cats

#11

Post by bobwaycott »

minimaxir wrote: post #4
danso wrote: post #2

The content of this article is pretty cool...but if I had to come up with a HN title that a deep learning process would surmise would do extremely well, it would be literally this. Or maybe "Show HN: How to use deep learning to optimize honey production from beehives" edit: The title of the submission has been changed; originally, it was something like "Using Deep Learning to Keep Cats off the Lawn"

Although incredibly off topic (the Cats = upvotes meme isn't even true on Reddit), a submission title's phrasing is not the primary cause of a HN story's success/failure. (and the proposed Show HN would make me raise an eyebrow in thinking the post is linkbait and cause me to flag it)

Why would you flag something based on the title alone? That seems to be an abuse of the flagging privilege.

Re: Chasing Cats

#14

Post by Faaak »

emeraldd wrote: post #9
aab0 wrote: post #7

How necessary is the deep model there? It seems like a simple motion detector would work just as well since he doesn't mention using the lawn himself.

Looks like there is a sidewalk through the lawn. Which probably means using a motion detector would force him to sacrifice using the front door.

You could either:

1) set the motion detection area only on the lawn 2) Set a delay for like > 30s movement 3) Set times of movement (ie. when they're at work)

But the big question is: did it work ?

Re: Chasing Cats

#15

Post by dharma1 »

aab0 wrote: post #7

How necessary is the deep model there? It seems like a simple motion detector would work just as well since he doesn't mention using the lawn himself.

Semantic segmentation/FCN isn't necessary since the spray isn't targeting the cat location specifically - you could just use a whole image classifier. You don't need a TX1 either, you could run this on a spare phone

https://www.tensorflow.org/mobile.html

http://mxnet.readthedocs.io/en/latest/how_to/smart_device.ht...

https://github.com/jetpacapp/DeepBeliefSDK

Re: Chasing Cats

#16

Post by minimaxir »

bobwaycott wrote: post #11
minimaxir wrote: post #4

Earlier quoted context omitted.

Although incredibly off topic (the Cats = upvotes meme isn't even true on Reddit), a submission title's phrasing is not the primary cause of a HN story's success/failure. (and the proposed Show HN would make me raise an eyebrow in thinking the post is linkbait and cause me to flag it)

Why would you flag something based on the title alone? That seems to be an abuse of the flagging privilege.

In my experience, there is a high correlation between linkbaity titles and spammy content. (I can unflag if necessary, but that has been rare)

The official rule is to flag if the content is spammy/off-topic.

Re: Chasing Cats

#18

Post by reitanqild »

minimaxir wrote: post #16
bobwaycott wrote: post #11

Earlier quoted context omitted.

Why would you flag something based on the title alone? That seems to be an abuse of the flagging privilege.

In my experience, there is a high correlation between linkbaity titles and spammy content. (I can unflag if necessary, but that has been rare) The official rule is to flag if the content is spammy/off-topic.

Did I read you correctly?

Flagging first without reading in deed seems like abuse of flags. (Possible exceptions for very obvious offenders against site guidelines.)

Re: Chasing Cats

#19

Post by RickS »

This is great. There's a project that's the inverse as well: http://www.quantumpicture.com/Flo_Control/flo_control.htm

A door that only lets cats in, based on image recognition.

As an aside, I've really enjoyed the particle photon so far. It was a little wonky at first when they didn't have persistent storage of state changes, but now that's up and running, it's flawless. It runs the lights in my house (via a relay just like OP) and has recovered from a few power outages with no attention necessary from me.

Re: Chasing Cats

#20

Post by danso »

minimaxir wrote: post #4
danso wrote: post #2

The content of this article is pretty cool...but if I had to come up with a HN title that a deep learning process would surmise would do extremely well, it would be literally this. Or maybe "Show HN: How to use deep learning to optimize honey production from beehives" edit: The title of the submission has been changed; originally, it was something like "Using Deep Learning to Keep Cats off the Lawn"

Although incredibly off topic (the Cats = upvotes meme isn't even true on Reddit), a submission title's phrasing is not the primary cause of a HN story's success/failure. (and the proposed Show HN would make me raise an eyebrow in thinking the post is linkbait and cause me to flag it)

Ah I meant less that the title itself would get upvotes, and more that the title reflects the kind of content popular on HN. Deep learning, of course. And cats, though a quick search on Algolia reveals fewer 100+ upvoted cat stories than I would anecdatally assume. Would love someday to see a survey of cat vs. dog owners among HN users (though it'd be heavily biased towards urban dwellers, e.g. SF and NY, in which cats might be the more practical choice)
Post reply on HN