I'm cracking 50, and I can't even imagine "retiring". In addition to my dayjob (which is plenty rewarding) I also volunteer for FEMA, Skywarn, ARES, and ARRL. Also looking to get my EMT cert. If I sat around on some beach somewhere drinking Pina Coladas I would probably lose my freekin' mind out of boredom.
I hate the term "AI" (even though I am CTO of a company with "AI" in its name, but since we use machine learning/DCNNs in our systems, it’s very trendy). The problem with "AI" is the "intelligence" part. Intelligence is a construct like "porn", like in the famous words of Justice Stewart about defining the latter "...but I know it when I see it". At best, it's very ambiguous -- and misleading at worst. They have been many attempts to quantitatively and qualitatively define intelligence -- none of which I find particularly satisfying and neither do any three given scientists in a room agree on a single interpretation. My problem with TFA is that it is comparing apples to oranges; deep convolutional networks are very different tools useful for a subset of problems than the ones using Bayesian inference and other statistical methods. Brute force methods like image morphology, object counting, and transforms are useful for even yet an entire Other set of problems. To say that one has displaced another is an error, in fact in most useful, modern, production systems a combination of all three is utilized, each to their purpose. To make direct comparisons between them while implying the historical decisions to use one or the other are due to Moore's Law is a false equivalence.
I took five years of Spanish -- high school and university (I thought I'd since forgotten it, this was 30 years ago). A few years ago I found myself in Barcelona for a week. First day, I was completely lost. By the end of my stay I found it had come back to me, at least for rudimentary conversation. I even picked up a little Catalan. This reminds me of a conversation I had with a former boss. He asked if I spoke Italian. I said no. "But you speak some Spanish, right?" To wit I sheepishly said, "yeah but it's been a long time". Then he rattles off something in Italian and asks "What did I just ask?" .. "Something about wanting 4 glasses of Tignanello wine for the table" And he says: "See? You have enough to be in the conversation, or at least listen. Talking is overrated, anyway..."
I'm CTO of a machine learning/computer vision company, and quite a bit of the team's work is "gluing libraries together" -- but even so, it's an intensely interesting, rapidly evolving field. Plus I get to play with pixels and cameras and VR. That being said, I paid my dues in the 90's and early 2000's doing web services and boring J2EE and gluing Java to native and PHP to Perl to CGI [bleech]
How did you earned that position despite your experience? Seems impressive to jump that far given that industry prefers very young and buzzword oriented (or phds in that particular field).
All and all a good intro tutorial that gets into some of the common professional use cases. On "constant bitrate" assumptions and some of the subsequent discussion here, ANY transform-and-entropy codec like VP9 or H264 will ALWAYS be variable bitrate internally. In the pro broadcast world, where you can't go over your FCC-allocated radio frequency bandwidth allocation by even one iota (or nastygrams and fines ensue), this is "handled" by actually having the encoder understoot the actual target (say it's 5mbit), and then the stream is padded with null packets to get a nice even 5mbit/s. This also happens with LTE broadcast as well. The encoders that do this reliably are fabulously expensive and of a rackmount flavor.
Reading this reminds me of all of the days spent using Manzanita MPEG software to get streams to "work". While its true that the bit rate may/will fluctuate, it was the muxing software that saves the stream. The muxer would introduce those null packets. The studio I used to write automated workflows for would get work specifically because we could make streams work that other facilities could not. Manzanita software was always the difference. Rarely, did an encoding software/hardware output work directly. However, remuxing the same elementary streams with Manz software would pretty much always solve the problem. If that didn't do it, it was probably a VBV buffer under/overflow issue around a scene change somewhere in the stream.
I'd love to hear more details about those things. I'm guessing it's not as simple as wiggling the quality around to keep the output within a certain size.
That's basically it. All of the digital broadcast streams (over the air, cable, satellite) use a fixed bitrate per physical channel, and send an mpeg transport stream. A transport stream is built of fixed length packets. Within the transport stream you can multiplex different programs. OTA gets 20Mbps per channel, if you use that for one program, it's likely you may not use the whole thing, so you include null packets as needed to fill the stream. If you send multiple programs, you probably will fill the stream, so you have to reduce quality and/or play games with timing of i-frames and possibly adjusting program start times or commercial break lengths to avoid having multiple programs needing high bitrate at the same time.
I bought a Chinese 1080p action cam for $30 about 3 years ago... is there some kind of competition issue in this market? (and it came with the waterproof case and various mounts)
I'd say there's a competition problem in that there are too many very poorly made "me-too's" that aren't even worth the 30 bucks. The Yi's come pretty close quality-wise to GoPro, have excellent build quality -- and frankly I like the Yi optics better than GoPro's... but they ain't 30 bucks. Disclaimer: I used to design camera systems and used to work in Hollywood so I can tell the s* from the Shinola, so I consider myself fairly picky.
I'm fairly picky too, I tend to compare everything to my Nikon D5100 (maybe not as picky as Q. Tarantino). This $30 action camera has decent image quality but the physical build feels cheap... but it hasn't broke yet.
So, I wonder how many ad placements were made by these companies that specifically targeted experienced talent OVER 40 years of age? I'm going to guess a big, fat, zero.
I used to work for a smallish (<50 headcount) company that sold high-value software with a LOT of secret sauce in it (image processing algorithms) mostly to telco, wireless, and MSOs. The primary reason those folks wanted the source code was as insurance that if, as a smallish company, you went out of business, they would be jolly-well-rogered if they had deployed your stuff in the middle of their mission-critical revenue-generating operations. Our solution was always to use a mutually agreed third-party source code escrow service such as Iron Mountain. The only escrow release triggers were company insolvency/bankruptcy, or refusal to meet the SLA of support requests for an extended period of time (the -Off Clause).
you could also send the source code to a third party notary to send to them incase you went bankrupt and also release your product open source when you go bust.
