|
The only issue for us is the size of the models so we'll have to work out a simple way to allow you to pick and choose what, if anything, comes down the wire when you install. Still: this opens up a fun world of opportunities for us.
New release first thing next week.
cheers
Chris Maunder
|
|
|
|
|
Blue Iris Software has just announced the release of their first version that uses senseAI as their AI provider. This is very cool.
On top of that we've had word on senseAI and Agent DVR integration.
We'll be needing Custom model support and GPU support to ensure we can provide their end-users with the best experience possible. The timeline is within 2 weeks at the outside. Who needs sleep.
cheers
Chris Maunder
|
|
|
|
|
According to a recent StackOverflow survey 45% of developers use Windows, 25% macOS, and 71% use Visual Studio Code.
I've been trying, for years, to make a mac my primary machine and I've gone to many painful and frustrating lengths to get there. Except I've never quite got there.
I'm happy to announce, though, that as of tonight we have the senseAI development experience on Windows and macOS essentially on par (give or take the .NET WinForms demo app). If you live in VS Code you won't see any difference.
The caveat here is that we're talking macOS on Intel. macOS on Apple Silicon is a whole different bag of broken parts and we're still waiting for libraries such as the ONNX runtime to support Apple Silicon, and we still have issues with some .NET image libraries on the new chips.
Things will progress and with the release of the new M2 chips I'm sure everything will catch up sooner rather than later. It was announced just a few days ago that even Visual Studio 2022 would run natively on Arm64, meaning Visual Studio running in parallels on a new mac may now be an option for the times when one wants a little arm-chair style comfort in their development day.
cheers
Chris Maunder
|
|
|
|
|
We've just released senseAI v1.4.0 and we fell into the trap of "just one more thing" while working to get the release out.
Here's a timeline you've probably had yourself:
- We need to add feature X. Let's go!
- We hop on a Zoom call and start brainstorming ideas on how to do this.
- We settle on an idea and go for it
- One of us, deep in the bowels of the code, realises there's refactoring that could be done
- To release, then refactor, then re-release? Or refactor, since I'm here anyway, and then release?
And you know what happened. The urge to release the best code beat the urge to provide our users with new functionality.
Maybe we've been burned too many times by rushed releases. Visual Studio seems to be less and less stable and more and more bloated. Apple's iOS has an exciting history of chaos. CodeProject itself has broken things left, right, and centre more times than I care to remember.
But here's the issue with waiting till it's perfect.
It will never be perfect. We'll be lucky if it will be stable and useable. If we're really, really lucky it will be useful. That's simply the reality of software development.
So by holding off on releasing the code we ended up making the engine run smoother, a little faster, a little more efficient, but (to stretch the analogy) we're not letting anyone drive a perfectly functional car. And the purpose of writing software is to provide something your users can use. Hence the term. Users.
Having said here's a little more background to our decision.
The change we made, instead of releasing the code, was in the manner in which you add a module. One of the core pieces of senseAI is to make it easy for developers to integrate AI into their apps by taking wonderful Open Source projects and aggregating them into senseAI's environment. We make sure the runtime is sorted out, the packages are installed, the API is solid, the installer works, and developers just provide the fun stuff.
Our change was that we took a process that required maybe a little too much understanding of the mechanics of how it all works, and turned it into a process that required writing a single callback function. 240+ lines of reasonably complex code down to 65 lines of very straightforward code.
So our decision was wasn't "do we release new functionality" but rather "do we continue to encourage developers to use a painful way of adding modules when we know full well that in a week we'll have a much, much better way".
It's these sorts of decisions that have to be weighed against your users, your marketing objectives, your budget and your stakeholders. Maybe marketing needs something, anything to talk about. Maybe Marketing can't handle the complaints when poor release decisions are made. Maybe you need to demonstrate progress or maybe you need to know whether it's even worth continuing in a certain direction before even thinking about refactoring.
Development time is super, crazy expensive, so these decisions are important.
cheers
Chris Maunder
|
|
|
|
|
We're releasing a new version of SenseAI Server today and unfortunately we're still on Wix as the installer. Maybe it's grown on us, maybe it's just beaten us into submission.
A few highlights of this week
- When attempting to get a code signing certificate make sure your company's info is up to date. We've moved offices and our old address is everywhere. Our new one is, too, but unfortunately you don't get to pick and choose which sources of The Truth a cert vendor will use. Update your D&B info, and do it over the phone (via the web is all but useless if you're not a US company)
- When upgrading to Visual Studio 2022 don't burn your bridges. Our old webforms app didn't load. It didn't just "not load" it caused VS 2022 to throw an arithmetic overload exception. That's a bit dodgy. The issue?
<IISExpressSSLPort />
Evidently the project loaded was parsing the value under the IISExpressSSLPort node and must have done a int.Parse instead of a int.TryParse.
The other frustration is that all our menu, toolbar and environment customisations were not brought forward from 2019. Nor were many extensions. It's understandable that some would not be compatible (or needed) in 2022, but to not warn or offer compatible versions means a lot of re-setting up of your environment
- We've moved to .NET 6 entirely for SenseAI. .NET 5 has barely cooled on the racks but it's already so last week, so to make things neater for our dev work in Linux and macOS via VS Code, it's .NET 6 all the way. Which isn't supported by VS 2019, for no reason I can fathom. Well, apart from lots of marketing reasons, but actual, technical reasons that can't be surmounted? Nope, can't actually think of any, but then again I don't write IDEs. I just pay for them
cheers
Chris Maunder
|
|
|
|
|
We've officially given up on WiX. We tried, and succeeded, but it was not worth the pain, or the complexity, or the having to enable .NET 3.5.1, or the feeling that it's "by engineers, for engineers". We're engineers, and we (collectively, our entire industry) need to stop assuming that our end users have the time, patience, interest or understanding to deal with our output.
As an exercise I'm stepping back and looking at what we present to our users, and looking at each decision we're asking a user to make. The question on my mind is "Does making them choose improve the experience" and I'm learning that no, it often doesn't.
Providing a choice is often a way for a software developer to avoid making a choice themselves.
I'm not talking clear functional choices such as "Do you want coffee or tea?". I'm talking choices such as whether a message is entered as Markdown or HTML, whether you want a two column or three column layout, whether you want to use the built in Iris GPU or that Nvidia Beast 3000 I see over there. Makee thee call, based on the reason you wrote the app, and focus on backing that decision to the extent of your abilities.
If you're running around in circles adding code to satisfy 1% of your users, make sure those 1% of users are fundamentally important to the survival of your code, and not merely the 1% who contribute 90% of the complaints.
If you're also providing a choice simply because it's possible to provide a choice ask yourself whether that choice even makes sense. "Do you want this app to run smoothly and efficiently or do you want the version with 1 arm tied behind its back?" I'm sure there's a use case, somewhere, for the second option, but do you have the capacity to continue to support that use case at the expense of dev time on the main use case?
So for SenseAI we're switching to an installer that allows us to get more done with less fuss and with a far more gradual learning curve. Our SensaAI code is all open so our decisions around what installer we choose impacts how hard it is for someone to work with the codebase as a whole.
The less decisions we ask you, our end users and collaborators, to make means more time for you to explore and build.
cheers
Chris Maunder
|
|
|
|
|
We're finishing up our first version of an actual, real, grown up installer for SenseAI and it's been an interesting switch from deploying web applications to deploying an app on someone else's machine. Without being there. And trusting them to press the right button.
It's terrifying.
Here is our set of golden rules for installers
- It should be blindingly obvious where to download the installer.
We've all seen download pages that have "DOWNLOAD" ads strewn across them. That's just juvenile marketing and results in anger. We've also all seen (and I'm looking at you Python and Microsoft) pages that list 15 versions. And another couple of links for long term support, latest build, legacy support, current version, recommended version, and standalone version. It's like they don't actually want you to download.
So as few options as possible, preferably just one.
- The installer itself should not raise difficult questions.
Do you want to run this as a low priority background task on startup, or a Windows service? Should it be installed globally or locally? When installing should we use your built in Intel Iris graphics, or use the NVidia CUDA extensions or do you have an AMD graphics card with at least 4Gb RAM? I can imagine the look my father would be giving the installer.
Either reduce the questions to what you know is possible, or phrase them in sensible terms: "Do you want to save time or save money"?
- Allow the install to happen offline
When I travel home to see the family in Australia I'm usually saddled with terrible internet. 1Mb/sec with latency in the order of hours, it seems. I always update my laptop and phone before I leave and download anything I might need to install onto a thumbdrive just in case.
Alternatively, say you're installing something onto 3 machines and really don't want to (or can't) download 500Mb three times. If at all possible, provide at least the option of downloading the lot.
- Allow the install to be a minimum install
We at CodeProject generally prefer to suffer so our users don't have to. A little extra work our end in creating an installer that provides the means to choose what you wish to install before you download the big chunks is worth it. The Visual Studio installer, at 1Mb, is brilliant. Downloads in a second and then offers you the means to gorge on 20+ Gb of downloads, or something much leaner for those who don't need their system to be a cluttered basement.
- Provide all the features of an installer people expect. Progressing installation for when you want to add bits you didn't install previously. Updates when a new version comes out. Repairs for when your users accidently break stuff. And, of course, an uninstaller. An uninstaller that uninstalls cleanly.
And having said this our first installer won't have all these features. Progressive installs will be in the next version, and we're still debating on whether our first iteration will be a totally offline installer or a bootstrapper that pulls down the parts as needed. Again, both options will be in place as soon as possible.
Think about who will be installing your software and assume they don't understand the terms you use, don't have any time, and don't actually care. Software is a tool for doing other things. An installer is a tool for making something else that will do other things.
Our job is to allow people to do awesome stuff that may happen to include some of our tools at some point. They have better things to do than try and read our minds and/or become experts in our esoterica.
cheers
Chris Maunder
|
|
|
|
|
This is way overdue, but we've released CodeProject SenseAI Server.[^], Beta 1. Beta 2 is coming any day now, depending on Matthew's swearing and the amount of sleep I can get.
To back up a little:
At CodeProject we've been forever fascinated with the possibilities of AI. Dave has forever been the type of guy who will build, regardless of the technical obstacles, solutions that will make his life better. Matthew is up for anything as long as it involves learning and sharp, bleeding edges, and I'm addicted to data processing. Ryan (who is now Senior Content Strategist at ContentLab has a rich history in AI development and spent forever trying to get us up to speed with the latest trends and technologies in AI.
And we all found it insanely frustrating that at every step it was a mess of Terminal windows, hunting installers, reading the fine print, tweaking config files, ensuring you had just the right version of Python, except when you needed the other version of Python, and don't even get me started on library dependencies.
It often ended poorly so and we decided enough was enough and we would commit to creating a system that would allow the average developer - the one who's way too busy, with way too many things on their plate - to explore and have fun with and integrate AI into their programming life with zero fuss.
et voila.
SenseAI is a standalone server you can install anywhere. (caveat: "anywhere" currently means a Windows machine, but with Release 1, due in a week or so, it'll be Windows, Linux, macOS, and anything that will take our Docker image). The code is all there for you to grab and dig through, and will build and run under VS Code.
How it works is you install it on your machine, start it up, and you have a simple little AI server that will happily run analysis on any image you throw at it using a simple REST API. We've included a very simple HTML page that demo's the current features. Face detection and recognition, scene detection and object detection are the current capabilities, but we'll be continually adding new features as we find them.
And "find" is the important point here. Our goal isn't to write AI modules from scratch. There are way too many people far smarter than us who are very good at that stuff. Our goal is to aggregate the best open source projects, wrap it all up in a safe, tested, consistent, not-breaking-because-you-forgot-an-environment-variable, documented and supported service with a single API.
The architecture is simple: There's a front end server that handles the API calls. This is a lightweight .NET app. The backend is made up of a collection of independent modules. These modules typically handle one AI task (eg object detection, sound analysis, etc). When you send a request to the server that request finds it way to the correct module, which processes it, and sends the result back to the server, and from there back to the caller of the Server's API. Each module is totally independent, can be written in any language that can run in the environment, and can use it's own AI models. As the system evolves and we come across modules that are better, faster, smaller, or more accurate, we'll swap out the previous modules for the updated modules with no change in the API. You'll just see a little notification on your server's dashboard letting you know there is new goodness to be downloaded.
We'll be publishing examples of modules we've built and added to demonstrate how we'll be including new modules, and we'll also be providing a developer SDK in multiple languages to make communication between the included modules and the server simple.
We're starting small, we have lots of rough edges, but we're committed to building something awesome here. We hope you can join us and help out.
cheers
Chris Maunder
|
|
|
|
|
Each year we tally the reputation points earned by those who spend their time answering questions in Quick Answers and the forums, and by those who share their code and hard fought knowledge through articles and tips.
It's these developers who make this community what it is and each year I am again humbled at the generosity, the depth of talent, and the endless patience in helping those new to the craft, or just new to a particular technology.
A heartfelt thanks go to our new batch of CodeProject MVPs for 2022. The Most Valuable Experts for helping answer questions, and our Most Valuable Authors for their incredible articles.
Congratulations, and thank you, to the following:
| Most Valuable Expert | Most Valuable Author | | #realJSOP | #realJSOP | | Afzaal Ahmad Zeeshan | Abdulkader Helwan | | BillWoodruff | Christ Kennedy | | CHill60 | gggustafson | | Chris Copeland | Glenn Prince | | Christian Graus | Han Bo Sun | | CPallini | honey the codewitch | | Dave Kreskowiak | Jarek Szczegielniak | | Gerry Schmitz | Joel Ivory Johnson | | Greg Utas | Kenneth Haugland | | k5054 | Marc Clifton | | KarstenK | Marcelo Ricardo de Oliveira | | Luc Pattyn | Marijan Nikic | | Maciej Los | Matt Scarpino | | merano99 | Michael Haephrati | | OriginalGriff | Michael Sydney Balloni | | Patrice T | Mircea Neacsu | | phil.o | Mohammad Elsheimy | | Ralf Meier | Nick Polyak | | Richard Deeming | Pete O'Hanlon | | Richard MacCutchan | Peter Huber SG | | Rick York | raddevus | | RickZeeland | Raphael Mun | | Sandeep Mewara | Sergey Alexandrovich Kryukov | | Stefan_Lang | Sergey L. Gladkiy | | thatraja | Sergio Virahonda | | TheRealSteveJudge | Shao Voon Wong | | Tony Hill | Shaun C Curtis | | W∴ Balboos, GHB | Steffen Ploetz | | Wendelius | tugrulGtx |
cheers
Chris Maunder
|
|
|
|
|
As part of our Diving Into AI that we've been doing at CodeProject, we found ourselves with two issues.
- We didn't know Python anywhere near well enough
- We didn't like any of the current Python language references. And there are lots out there.
So we made our own! Sure, it's repeating, sort of, what's already out there, and sure, it was a convenient excuse for massive procrastination when we should have been doing real work, but it was fun, and we now have a reference that we like, and that suits our short attention spans.
Plus, we got to work with @Marc-Clifton on this who did lots of the heavy lifting.
Writing the reference also helped with #1. It forced us to understand the language at a much deeper level than merely reading examples, or formulaic tutorials. It also forces you into all the nooks and crannies of a language. And there are some odd nooks and crannies.
It's a work in progress, and presumably always will be. The URL is https://www.codeproject.com/ref/Python3, and please send us your suggestions, improvements, typos, rants, whatever.
cheers
Chris Maunder
|
|
|
|
|
It seems that the current state of AI is focused on the Scientists, not the Engineers. In other words, to use AI to do something like detect a package delivered to your front porch you need to understand Python, NumPy, Pandas, PyTorch, TorchVision, TensorFlow, Neural Networks and all the types of layers, Model Training, Transfer Learning, Calculus, Linear Regression, just to name a few.
To make matters worse, most of the books and articles I find seem to be geared to you getting advanced degrees in Math and Data Analysis.
I am an Engineer and Software Architect/Developer. I just want to pull in a package that either just works or helps me fall into the ‘Pit of Success’ while implementing the last mile. I do not want the users of my applications to have to suffer through complex and error prone installation and configuration. I do want easy configuration to tweak the application usage for specific optimizations for environments and use cases.
Granted, there are many repositories of Datasets, Neural Network implementations, Pre-Trained Models, but where is the tool that guides me in the selection of which to use for a particular use-case and hardware/OS combinations? Where is the package that provides or allows me easily to write a microservice for my Raspberry PI that, given a few if any pictures of raccoons, will signal my home automation system to turn on the sprinklers if a raccoon is on my garbage cans?
I have been looking at several open-source packages/repositories that attempt to tackle some aspect of this, but they are overly complex, difficult to build and test, huge, or lack extensibility.
As a result, I am looking at how I can leverage the vast collection of data, models, examples, and articles available and wrap up all the science into usable Lego® blocks to make using this amazing technology easy.
Stay tuned for ongoing progress reports, and the occasional rant, as we try and figure this out.
"Time flies like an arrow. Fruit flies like a banana."
|
|
|
|
|
I often say that JavaScript survived and dominated because it offend everyone less than its competitors did. It's ubiquitous. It runs everywhere. It's also a horrible mess of ridiculousness that sends developers insane.
But I loves JavaScript. It lets me be lazy and it lets me do totally stupid things and it forgives me and sometimes even works.
So speaking of ridiculous languages that thrive in spite of themselves, let's talk about Python. Python is like JavaScript in that it just kinda hung around trying to be useful in its own awkward way. It soon found itself being used as a scripting language, a teaching language, and a language you can slip on like a pair of sandals when you're rushed, with the same elegance and style therein. Type your code straight into the interpreter and bang, you're away.
This also made it great for Jupyter notebooks. And hacking together scripts to analyse data. And to perform operations on data. And to be the underpinning of the biggest paradigm change in computing, namely Artificial Intelligence.
See? The awkward introvert does come into his own eventually.
So we at CodeProject are finding ourselves surrounded by a sea of Python with no option but to hold our noses and jump right in.
As an ex-Environmental Engineer surrounded by FORTRAN writing physicists I still get the shivers when I see some of the Python code floating around. We can use more than 72 characters, guys, so please feel free not to name your variables x, y and z. But there's also a lot of amazing Python code out there. Incredible stuff.
This is the stuff we want to see more of so we're starting from basics and will be diving into Python, data analysis, and then some of the gnarlier Open Source Deep Learning and Artificial Intelligence packages available today. Our goal is to promote the knowledge of the field of AI but to do it by following the most treadworn path using Python. We're not trying to reinvent anything: rather, we want to present it back to the community in a form that's easily digested.
Right now we're most of the way through our initial offering which is the most basic Python Reference we could conceive. Short, code-first, simple examples, and an emphasis on explaining "why" instead of "what".
It'll be a work in progress but we hope to have the first version live before Halloween.
While you're waiting, make sure you be awesome when you code.
cheers
Chris Maunder
|
|
|
|
|
via Jeff Hadfield, who's at BUILD keeping an eye on things for The Code Project:
By now, and especially if you have been following @thecodeproject on Twitter, you’ve seen a lot of Tuesday’s Windows 8 developer preview news.
Microsoft billed this developer-focused launch as a “new opportunity for developers.” And that it is, but exactly what the opportunity is remains to be seen. To their credit, the Microsoft teams have certainly done a lot of work on building new UI foundations, making touch an equal partner with keyboard/mouse. And the new hardware is also impressive. But while the pieces are all there, many questions remain to be resolved between now and when Windows 8 finally ships (we’re guessing in about a year).
There’s plenty of coverage about the new UI, and a lot of it can be found on the official Windows sites. We’ve taken careful notes – and we’re happy to add any commentary if you’d like, but rather than rehash the #bldwin Twitter hashtag (re-hashtag?), a few words of perspective might be better.
Our friend David Platt reminded us of his recent MSDN Magazine article and emphasized his admonitions there – about making sure the war hasn’t moved to new fronts while Microsoft continues to fight the same battle in the same place. Has the market moved on? Maybe. Can Microsoft lead the market again? Maybe. The company has a strong history of letting others blaze the way and then stepping ahead, much like stepping in front of a parade already in process. (Which in turn reminds us of the final scenes of Animal House, but with less debauchery.)
Developer buzz is mixed – excitement about the opportunity, sure, but fears also about it being too late. Here’s the thing: Windows still has dominant market share worldwide. In today’s presentations, Microsoft claimed over 450 million Windows 7 licenses sold, making it – as of last Friday – more in use than Windows XP. That’s a lot of seats, and a lot of opportunity for line of business apps and consumer apps.
But many, including me, express some skepticism about how ubiquitous and affordable touch-based PCs will become in the short term. Will they be more than a niche product? Perhaps, but the iPad, for example, is still not affordable for most. Get a nice Windows 8 slate/convertible at the $400-$500 USD price point, though, by next fall, and you’ve got a winner.
Again: metric tonnes of great new features and tonnes of “goodness” included in the APIs and UI guidelines. We’re anxious to get to play with the bits. Unsure how they will play on today’s hardware systems – dev bits and the Windows 8 slates were not available for press, so we won’t be trying any apps on our own. Let us know how you fare.
Finally, a few useful articles:
Sorry Apple, Windows 8 ushers in the post-post-PC era[^] The point here? That a capable computing device should be all things to all people, not a dumbed-down device that’s hard to do “real work” on. Agreed: I like my iPad, but as an authoring tool it’s a train wreck.
My friend Sean Gallagher has a more cynical look[^], but wisely points out that the big news is that Web developers who know JavaScript and HTML are now on an equal footing with traditional .NET (C#/C++/VB) devs. Interesting, and at least one of my developer friends expressed some concern about how that will affect traditional Windows developers. As I see it, Microsoft had no choice but to try to co-opt these “Web developers” in order to increase market share. The overall worldwide growth in ‘developers’ in general is coming from areas outside of the traditional .NET stack – so this is really their only choice.
There’s also a nice post at the Windows Team blog that shows a bunch of the UI updates[^]. If you can’t spare the time to watch the keynote (two and a half hours, seriously), read this but grab a few minutes to see the UI in motion in one of the videos.
And let’s say you’ve downloaded the bits already[^]. What to do? Try our friend Scott Hanselman’s guide (you’ll put your eye out, no warranties express or implied, YMMV) to running Win 8 from a VHD[^].
As always, let us know your experience, and post your tips and how-tos on the site.
|
|
|
|
|
We were at Microsoft's BUILD conference keynote address today and had a chance to see (and tweet about) some of the features revealed for Windows 8. Here's a rundown of the highlights:
Starting out, Steve Sinofsky discussed WIndows 8 on ARM processors, not that it's "equally at home on ARM and x86". Hardware graphics acceleration is "baked in" to the OS.
As you've probably already heard - it wasn't a secret going into BUILD - the Metro UI introduced with Windows Phone 7 is being adopted in Windows 8. The Metro-style even extends to the remote desktop app, which supports remote touch.
Of course, you're probably interested in the developer side of things. To paraphrase a tweet from @DavidJKelley: programming becomes a life style choice in Windows 8 - HTML5, Silverlight, C++/XAML its all good. The new version of Expression Blend will now work with HTML5/CSS as well as XAML, and they demo'd converting a Silverlight app to a Win8 app by just changing some using statements.
Developers also now have acces to SkyDrive access just like local file system through APIs.
There's a lot more, and we'll be covering much of it in the Daily Insider[^] tomorrow, so make sure you're subscribed.
Finally, if you're ready to play, you can get your Widows 8 dev preview tonight at dev.windows.com. No activation needed (though no support provided, either).
|
|
|
|
|
Since The Code Project's privacy policies comply with the strict eTrust rules, we are now eTrust certified. This puts a strong emphasis on the fact that, when you become a member of The Code Project or subscribe to one of our email newsletters, you know how we use your presonal information and can trust we follow our rules and the guidelines established by eTrust.
You can read more about our privacy policy at http://www.codeproject.com/info/privacy.aspx. To learn more about eTrust and its privacy requirements see http://www.etrust.org/.
|
|
|
|
|
Over the next few weeks we're going to be posting a series of blogs outlining what we're up to (and the how and why) to give you more insight into what goes on behind the large orange doors of CodeProject central.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
