This should be another fascinating evening. Should I bring my Roomba vacuum cleaning robot? Will my wife let me take it away? Is Augmented Reality the same as Virtual Reality? A quick Google of WikiPedia produced " An AR is an interactive experience of a real world environment whereby the objects that reside in the real world are augmented by computer generated perceptual information, sometimes across multiple sensory modalities, including visual, auditory, haptic somatosensory and olfactory. " Wow!
( haptic = tactile, and somatosensory = a complex neurological system including touch, balance, body positions, joints etc.) Wow again!
An easier definition from techopedia is
Augmented reality (AR) is a type of interactive reality based display environment that takes the capabilities of computer generated display, sound, text and effects to enhance the user's real world experience. That sounds better.
I was only able to attend a short part of this evening so I have had to work off the audio recording so apologies in advance if I have misunderstood the proceedings.
Sue began by giving a very clear presentation of her interest in Mandelbrot pictures and fractals. Nature is full of self similarity. Simple direct media layer (SDL) (libsdl.org ) she found bindings for many languages but she likes to use the Go Language because it is very fast and compiles and runs on the fly. However it is not thread safe.
You create a window of the size and position you want and run the Mandelbrot set for the size of window. You call a Present method in order to get the plot. She likes trying different colours. She presented several interesting pictures.
Rob Pike (Bell labs) invented Go which is like C, but it is easier and is garbage collected and does memory management. It is strongly typed and requires precise formatting. It can be used a lot like a dynamic language because it compiles so fast. Go is also very well structured and there is generally only one clear way to do things and so code is clear and does not allow people to be too creative and obscure. Many software companies now will only employ people who can write in Go.
Peter then give a short presentation on his use of Python 3 for producing contour maps from point height data on his Bowling green. He had been looking for some time for some software which would do this and found a Python library called matplotlib which did a neat job. He had to do some work in writing the Python code to read from a data file since most demonstrations of plotting auto generate the data points in a regular grid and he found it was not too straightforward to make it read an input file instead. The file had been created originally some years ago by using a horizontal scanning laser and a pole with a sensor on to move every 2 meters across a bowling green. The software I used does not require a Cartesian grid of data points (heights) which makes it much more general.
He was not able to show it running unfortunately as he had not used it for a while and needed to get the environment set up but had run out of time. He showed the original data set diagram and the results as some printouts of the contour diagrams with false colour representations. He promised to show the plots running at a future date. He liked Python because it was very flexible and produced compact programs which were easy to follow and there was a lot of support from complex library routines. It could be used interactively line by line as well as as a module.
After Peter, Steve gave a very interesting presentation. After the talk Duncan had given on archiving digital photographs and Digkam in July, Steve had wondered about the tagging of images and keywords automatically using Yolo. (You Only Look Once) this is an image identification system using neural nets to pick out and identify objects in an image. He thought it would be possible to use it for this purpose although you might need a good body of photos with keywords already defined manually in order to train the neural nets. Otherwise there could be a lot of errors in identification. The Yolo v3 software apparently claims to run extremely fast. Although Steve reckons you need multiprocessor hardware because his runs took about 10 sec which he considers too long.
[Ed. I don't think we have ever had a discussion about neural nets before.]
Steve copied and downloaded the Neural Net program Yolo and used an example set of images given by the author of the website. He wrote a shell script (bash) with one argument of the filename with the images. He redirects errors to /dev/null and then tails the last two lines of the output file passing it to awk with : which effectively produces the keywords. He also echoed the keywords as they were produced in order to see where it was in the list of images. He found the most difficult part was to find the correct way to tag the images, he discovered the ITPC Keyword was the correct way. It can determine false presence of an object. The software shows a second image derived from the first with boxes drawn around objects it recognises with the keyword associated with it. He tried some examples it was very good at recognising cats, dogs, aeroplanes and sheep. But thought a church notice was a microwave and a fridge was a noticeboard. It "thought" a Commodore PET was a laptop on a dining table. [ It all sounds very impressive to me. Ed.]
There were no other formal presentations as several people said they did not develop software. However there was quite a long discussion about older software and systems like BBC Basic or Microsoft Simple Basic and some on reverse engineering to correct faults in compiled software for which the original code had been lost. [Ed. Not for breaking copyright note.]
Paul had brought along old books on software including Archimedes Assembly code which triggered a few reminiscences.
I am sorry I missed most of this evening and would like to have a go at this myself. I saw the My Little Pony pictures on the website!
I was not sure which version of Yolo Steve had used as the author claims v3 is faster than other software.
At the Committee meeting we decided on the following list of future events, but as always, if someone else would like us to introduce some other topics we would be delighted to accommodate them.
|ICENI Future programme 2018/2019|
|October 17th||Robotics and Augmented Reality||All|
|November 21st||Slide/Video Evening||All|
|December 19th||Gadgets and Party Evening||All|
|January 16th||Winter Social Evening - TBA||All|
|February 20th||NAS + RAID + storage solutions||All|
|March 20th||New Devices and hardware SSD||All|
|April 17th||Mac Evening||All|
|May 15th||AGM - Computer Surgery||All|
The Social evenings dates and venues are just suggestions at this stage.
Our meetings are held at the Bourne Vale Social Club, Halifax Road, Ipswich IP2 8RE , for a map and other details please see the website.
Membership fee currently £15, visitors free.
"ICENI does not have any Insurance cover for computers or other equipment so please be advised that you bring machines to the club at your own risk."
However many household insurance policies will include cover away from home often with no increase in premium. (Ed.)