I worked for a company in 2000-2001 that had a small branch that built custom industrial machine vision systems. It was probably the most fun I've ever had in my career. Anyway, one day the head of the branch gathered everyone together and talked about how he wanted to break off from the master company and build a machine vision startup. Everyone was extremely excited and most agreed to follow him on the spot.
One week later 9/11 happened.
One week after that the same branch head gathered everyone together and laid off 80% of the staff. You see, "industrial machine vision was no longer viable"[1]. We were crushed and I moved on to other fields.
I haven't though about machine vision much in the past 10 years, but I think I'll read this book. Thanks for the submission.
[1]: Vision system to seek and destroy on the other hand...
I worked in industrial machine vision from about 1990-2005. What we saw happening was a gradual shift from dedicated systems with DSP and specialized hardware to PC framegrabbers with most processing on the host to firewire and cameralink cameras with everything on the host. The libraries from Cognex and Matrox as well as open-source solutions have become so good and easy to use that some of the mystique of machine vision has disappeared.
I think there are still tons of interesting machine vision applications but the systems development itself has become largely commoditized.
Eh? Strange. 9/11 was, as terrible a tragedy as it was, great for computer vision. DHS and DOD were throwing money left and right at researchers and companies (through SBIRs), though mostly for surveillance, activity recognition, vehicle recognition, not industrial machine vision, as far as I'm aware.
I wouldn't expect a large company to pivot, necessarily, but if the branch head wanted to do a startup, he probably missed out on a huge opportunity.
Hi - I'm the author. If you can't download, just wait a day or two - there aren't so many people interested in this that the server will be struggling for ever. Please don't torrent it as I update it every few days and I'd rather not have redundant old versions full of unnecessary mistakes out there.
I still can't get the full download, even after the update. I've been trying all day. It gives it a good try, but cuts off about 1/3rd of the way through. I'm still very much wanting a copy.
I think it is just because it is on the front page of HN. I had more hits today than in the last year! Just wait a couple of days. It's not going anywhere and there hasn't historically been a problem with downloading it.
I think it just has a time-out which is quite short (it's not my server it's mounted on). When the traffic is heavy this is activated before the download completes. I have managed to compress it to about half size now, and I'm uploading that at the moment. It takes about an hour because Rogers pointlessly throttle my upload speed at home.
the author posted this shortly after your comment:
" I've taken it down for a little while while I upload the most recent version that has a lot of small errors corrected. It will be back up in a couple of hours I expect. It's worth the wait..."
So it doesn't seem it's a bandwidth issue so much as a quality issue.
The download is huge because I made the figures in Inkscape and it can't produce compressed PDFs. If anyone knows how to solve this problem, I'd be happy to hear about it. Mail me at s.prince@cs.ucl.ac.uk.
I worked for a company in 2000-2001 that had a small branch that built custom industrial machine vision systems. It was probably the most fun I've ever had in my career. Anyway, one day the head of the branch gathered everyone together and talked about how he wanted to break off from the master company and build a machine vision startup. Everyone was extremely excited and most agreed to follow him on the spot.
One week later 9/11 happened.
One week after that the same branch head gathered everyone together and laid off 80% of the staff. You see, "industrial machine vision was no longer viable"[1]. We were crushed and I moved on to other fields.
I haven't though about machine vision much in the past 10 years, but I think I'll read this book. Thanks for the submission.
[1]: Vision system to seek and destroy on the other hand...