Does anyone know if this dataset has already been pre-filtered by some computer vision algorithm to weed out the uninteresting images? If not, is there any way to get hold of the data directly so we can process it offline?
317 karma · joined November 7, 2011
Does anyone know if this dataset has already been pre-filtered by some computer vision algorithm to weed out the uninteresting images? If not, is there any way to get hold of the data directly so we can process it offline?
If the ships travelled in groups of 3+ then in the event of a failure the other two ships could tow the failed ship? Is this feasible? I honestly have no idea.
[1] http://www.extremetech.com/extreme/175562-major-patent-expir...
;)
Similarly, are there any inexpensive but decent smart plug sockets available? (I should note that I'm in the UK, but I'm sure other readers would welcome US-specific solutions.)
And, same question for BLE tags - below ~£10 a tag would be awesome.
Now I run Sublime Text on a laptop that goes everywhere with me and deploy using a proper VCS. I would certainly use ShiftEdit again in future if I was stuck without my laptop and needed to make a few quick changes to something, but I wouldn't use it as a replacement for a local editor.
Of particular note is 'Flutter', a track where no two bars are the same, which was produced in response to the UK Government passing an 'anti-rave' law which attempted to classify dance/electronic music as 'repetitive beat music'. Party organisers were encouraged to have a copy of Flutter, a musicologist and a solicitor on hand in the event of the police trying to enforce these crazy laws. Fun times!
Firstly, maybe the cable is part of the antennae, thus negating this issue?
Secondly, maybe there's a heartbeart, the absence of which triggers an alert?
Finally, in practical terms, this probably won't become a problem until all bikes have upgraded. To deter a thief you don't need perfect security: you just need better security than the next potential victim.
A few suggestions to improve the writeup:
1. Consider going into more depth on the workings of the image processing techniques. The ones you have chosen are some of the more readily-accessible and, to the uninitiated, can be a delight to learn (esp the Hough Transform, imo). This will also help cement things in your mind and be useful when it comes to exam time.
2. Feature selection is obviously the most critical part of the process - consider going into more detail. (Un)Fortunately you got it right first time, so you couldn't document a process of trial, error, insight and finally success. In lieu of this you should at least be able to elucidate on the thought processes that led you to choose the features you did. If you are able to make a prediction that is subsequently realised then that demonstrates you are intelligently selecting features and understanding the consequences, rather than just getting lucky.
3. Your description of k-Nearest Neighbours is a little faulty - possibly the result of you trying to phrase what is essentially a simple concept in your own words? The user doesn't pre-define some set 'search radius' - the algorithm simply finds the k nearest neighbours of a given point. If your point is some outlier then the 'search radius' will be much larger then had you selected a more 'typical' point in the middle of a densely-populated area - for the same value of k.
Currently using a hi(ish, not Retina) PPI laptop with a standard monitor and, well, you know the rest.
IMO, their three suggested use-cases are quite flimsy.
"...enable surgeons to take notes during an operation without removing their gloves" - They already can, it's called dictation which is subsequently transcribed by a medical secretary whose time is far less expensive.
"check recipes while cooking without spilling flour on the keyboard" - touchless navigation of a pre-chosen recipe, maybe. Actually stopping cooking to search for a recipe/technique that you're unsure about? I would rather quickly wash my hands and use a more traditional input method to find what I want.
"search for TV programs without leaving their couch" - Given the range of the Leap device this really doesn't make sense - are they envisaging a wireless version you take with you to the couch/coffee table. Some sort of device that can 'remotely control' your television...
I still think Leap Motion is great technology, but still missing a decent application.
This is nit-picking if I'm right but potentially enlightening if I'm wrong (so apologies if it's the former): wouldn't /insomnia/ be the symptom of mania, and sleep deprivation be the consequence?
Also, I always thought of mania as being either very 'up' or very 'down' - is that my ignorance? I feel neither happy nor sad during these periods, just a lot more focussed.
I find my concentration improves tremendously after I've been awake for 20-24 hours, and I can continue being more productive than usual for up to about 40 hours, at which point the negative effects start to outweigh the positives.
Anyone experienced anything similar? I should note that I've not done this repeatedly/for extended periods and I only allow myself to do it if I'm sufficiently well-rested. (Otherwise I just figure it really isn't going to be worth it and I'm better getting myself to bed and getting an early start.)
I'm afraid I have no data to support my claims. In the absence of data I do have an anecdote though(!) - I remember when I was maybe 8-12 years old travelling long-haul and had been up for 45+ hours absolutely SMASHING a puzzle game (slide the blocks type thing) on my Gameboy - previously I'd struggled to get past the initial few levels. I'd not thought about it much until recently when I noticed this effect (end of Master's degree... you know the score).
Cleaners that CCTV themselves.
It could be done with something as simple as a low-cost smartphone worn around the neck, or in the future something more sophisticated like Google Glass.
Lots of potential issues (would the cleaners accept this condition? would clients be happy to have a video of their house sat on a server?) but it's a potential solution to the trust issues inherent in the business - two main ones being 'is my cleaner dipping into petty cash' and 'is my cleaner actually cleaning'.
Could the case be made that journalists are legitimate targets for investigation following on from Manning and Snowdon? My focus here is less on terrorism and more towards counter-intelligence. Were I a rival nation-state I would see an intelligence-gathering opportunity in what's happened: Rather than bribing/coercing people into become informants, there is now the option to invite them to 'blow the whistle' by posing as a journalist.
KSP was mentioned to Elon in some post-talk Q&A session - he hadn't heard of it and seemed unimpressed.
I've set to see a single one in the flesh. The only ones I've seen in photographs are tourist gimmicks. Granted, I've never travelled to North America, but I do live in a developed western country.
For me, that's one of the most significant things about the whole project. I'm no sandal-wearing Greenpeace-type, but I am convinced that if we're to continue trying to achieve developed-world quality of life for everyone on the planet then we need to start getting much smarter and more efficient with our energy usage. One way is 'stop living the way we do', the other way is this (and other advances like it).
Seriously though, I wish you the best of luck. My advice would be to get yourself a good portfolio assembled and then maybe approach small- to medium-sized companies and just be open and frank with them. If you can demonstrate you have the skills there will be someone out there who is happy just to have someone sat coding up briefs, reporting only to them. Larger companies might find it more difficult to get around the mindset that meetings are universally important and everyone eventually wants to be a manager. A small business owner might see someone like you as a godsend if you help them understand that in return for being largely undisturbed (in a social, obv not work sense!) you will be loyal and productive.
Obviously a Very Bad Thing, but if we feel it necessary to protect against that then we might as well write off transport altogether!
Anyway, enough speculation from me - I'll patient wait for Elon to finish his breakfast and press 'go' on the release.(Very frustrating!)
If the 'Loop is as I'm imagining it (based on scant details, granted) then the actual vehicle components are small, numerous and frequent. Given the short transit time it could be reasonably expected that people carry NO luggage whatsoever and could be quickly processed by a walk through a (potentially computer-operated) backscatter x-ray machine. Luggage runs could be carried in separate vehicles, spaced so that if anything were to 'go off' there would be loss of infrastructure but no loss of life.