hwoolery

You can look at MediaPipe's BlazeFace / BlazePose for faster inference for lower-end chips. I currently have an application that does realtime video with YoloV8-M on my Macbook at over 60fps

It's not total bullshit - compare Imatinib on GoodRX ($787) vs CostPlus ($14) for 30 tablets. What the company is doing is cutting out the (arbitrary) pharmacy markup. They take a 15% cut and are TRANSPARENT about their price (again, novel in the drug industry). If you followed their company, you'd also notice they have plans to further lower the margin by manufacturing their own drugs, but these things take time.

For sure, agree!

Let me add a bit of advice to any homeowners out there: learn to do the job yourself! I bought my house 5 years ago and I went from zero knowledge to remodeling the entire thing mostly on my own. Not only does it save you the hassle of having to deal with contractors, but in the cases when you need one, you'll be a lot more prepared to gauge the quote and timing.

"a few more remote friendly companies". Respectfully disagree - the market will make this decision, and now that employees have all tasted the benefit, they are going to work for companies that give them that freedom. On top of that, most companies have seen equal or greater productivity (no commute, fewer sick days, etc), and will at least downsize their real estate for cost savings. There's little argument to make people come back to an office if there is no obvious benefit.

I am switching to one (not yet delivered) for my iOS dev work from a 2017 MBP. My main reasoning is price, and since my laptop mostly just sits in clamshell mode. I got the 16GB one, but can't imagine the 8GB will suffer for development work. I still do all my machine learning work on a PC

Step 1. Farm out < one-month contract jobs Step 2. Profit

The cropping issue is an unsolvable one for AI. It doesn’t matter how unbiased you want the model to be, if an algorithm has to chose between 2+ faces with no “correct” answer, it can’t win.

You might just think, why doesn’t it just identify the skin color and crop randomly based on that? Now what if there is two races of men and one woman in the image? Or vice versa? You quickly get into if/else hell.

The algorithm can be tuned to no end, but in the end, it’s dealing with an infinite amount of possibilities (not just faces) and it’s a problem that proves Twitter just needs to let the users adjust the crop or focus themselves.

I agree -- on top of this, I think machine learning will play an important role. The power of machine learning + AR creates some interesting possibilities. You can check out our website for some examples https://www.2020cv.com

Apologies if the title is confusing. The demo does use CNNs, as mentioned to predict the 2D joint locations of the hand. The whole point of this system is that it relies only on a monocular camera, allowing it to run on a wide range of devices. CoreML on iOS and Tensorflow Lite on Android handle all the GPU inference on these models

That app in particular got sidelined for now, but we've released two others since then:

http://instasaber.com http://speakingpuppy.com

My project didn't get much attention on Show HN (or anywhere else), but I sent a cold email to Mark Cuban and offered to invest pretty quickly. So don't get too disheartened if your project gets a tepid response.

Orignal post: https://news.ycombinator.com/item?id=14459876

Yes, although wolves are making a comeback in some parts of the world, so there's something to be optimistic about!

Maybe in the future -- a bigger commitment to create the infrastructure needed for that. For now, just looking to get feedback and reactions : )

Gif of it in action: https://twitter.com/i/status/1032148934108409856

Hey HN, I made this app, let me know if you have any questions! It uses a TensorFlow/Keras model to detect the dog's facial features, and a custom AR interface for estimating 3D pose.

Assuming the miners still want to rent out the hardware and not sell it :) Even then, you'd have an instance of two parties both taking slices of the profit, which is hard to compete in a practical manner against the Amazon/Google

Serious question (as opposed to a snarky one): what happens when/if the crypto currency market takes a large downturn? If your pricing is based off using their cycles, Snark.ai would have to raise prices, right?

Of course it's a difficult problem, but why try and dissuade people from attempting it? Nothing novel ever came from somebody who didn't fail first : )

I would love it if I had the time and resources to do that, but hard to justify with how poor the retention was on iOS (understandably, from a gimmick like this). Quite a good chance I will be targeting Android first next app I do.

It's such a dumb idea it doesn't need much explaining, haha

Many people have requested dueling, but there's a few technical hurdles there : ) Nice work with your kid!

yes, that as well

Essentially, just using it to feed frames into the CoreML model, but not using the planar detection. However, I am using ARKit world orientation tracking in other stuff I'm working on like this: https://twitter.com/2020cv_inc/status/969628227130949633

I created an app that uses ARKit + Machine Learning (CoreML) called InstaSaber that projects a virtual lightsaber out of a piece of rolled up paper: http://instasaber.com/

It was more a proof-of-concept of the technology I'm working on, but I think it's worth checking out (it's free to download). I think you will start to see a lot more cool implementations of ARKit/ARCore that integrate with CNN's and do new and unique things.

Edit: 20k downloads in first week, I haven't bothered to market it though.

Uh, this is just a PR stunt for the movie at the bottom, right? The court report link is broken...

http://instasaber.com (app store: https://itunes.apple.com/us/app/instasaber/id1316793122?mt=8 ) InstaSaber is the most advanced augmented reality lightsaber app out there, using Machine Learning + ARKit. Yes, it is completely pointless, but I promise it's a few minutes of fun : ) . Free to download, $0.99 cents if you want to upgrade. info@2020cv.com

Uh, InstaSaber, of course: https://twitter.com/2020cv_inc/status/942849449193177088

If on Android, just hide inside your Tauntaun and pray... joking (we hope to get this running on Android soon too)

I hope to do a technical writeup of my process soon. In short, the training images were created from videos of rotating the tube around in hand at various angles and lighting conditions. These images were then annotated to determine the location of several keypoints around the paper. It generalizes pretty well, see here: https://twitter.com/2020cv_inc/status/945502116335906816

thanks! Still a lot of refinement left to do (this was initially released as a proof of concept). But I'm happy to see people from all over enjoying the gimmick : )