yeldarb

31 ms·

301.
by yeldarb 6y ago
It’s not entirely clear to me exactly what was meant by the grandparent post by converting to PyTorch or Tensorflow. There are two discrete concepts: training and inference. You can think of training a bit like the source code and inference
302.
by yeldarb 6y ago
I didn't clean it up and submit for app review but if you have XCode you can compile and run it by tweaking the sample app (start at 22:25 in the YouTube video above). Here are the weights: https://www.dropbox.com/s&#x2
303.
by yeldarb 6y ago
With the right training data, definitely. Check out the Oxford Pets dataset as a starting point (you can download in Create ML format from here): https://public.roboflow.com/object-detection/oxford-pets
304.
by yeldarb 6y ago
There are a handful of free tools you can use. We have write-ups of most of them in the "Labeling Your Images Yourself" section here: https://blog.roboflow.com/object-detection/ Roboflow (our startup) can imp
305.
by yeldarb 6y ago
Yep, in this example the weights are compiled into the app and inference is done locally on the iPhone processor.
306.
by yeldarb 6y ago
Apple's main goal with Create ML is to make something app developers can use to add machine learning powered features to their apps. So it makes sense that they'd want something that you can train with the hardware you already hav
307.
by yeldarb 6y ago
You can convert CoreML models to ONNX which is widely supported cross-platform for doing inference. https://github.com/onnx/onnxmltools
308.
by yeldarb 6y ago
We're actually running a benchmark of that very thing this week! I'll follow up with what we find.
309.
by yeldarb 6y ago
CreateML is getting really impressive. I just went through the process of using it to train an Object Detection model to detect sharks and fish at an aquarium (and deploying the model to an iPhone) last week. Final result: https:/&#x2
310.
by yeldarb 6y ago
I tried to fine-tune Deoldify (a neural net that colorizes old photos) to colorize Garfield comics a while back. The results after a couple hours of tinkering were promising. I still think it would be an interesting project to pursue. There
311.
The Challenge: Teach my phone to find sharks in images without writing any code(twitter.com)
5 points by yeldarb 6y ago|0 comments
312.
by yeldarb 6y ago
This video describes the process: https://www.youtube.com/watch?v=zvmhl-E4QQ0 Pretty clever.
313.
Designing for Augmented Reality(blog.roboflow.com)
3 points by yeldarb 6y ago|0 comments
314.
by yeldarb 6y ago
I saw some initial reports on twitter saying it was about the same speed as a 1080 Ti https://twitter.com/spurpura/status/1329168059647488000
315.
by yeldarb 6y ago
I had never heard of youtube-dl prior to this whole saga. But I used it for the first time yesterday and it works great. So maybe I should thank the RIAA for clueing me in!
316.
by yeldarb 6y ago
Is an empty .html file technically the shortest html quine?
317.
Epsilon: Generate and debug Python code with GPT-3(epsilon.shreenabh.com)
2 points by yeldarb 6y ago|1 comments
318.
The crazy story of how we got our .com domain(blog.roboflow.com)
5 points by yeldarb 6y ago|2 comments
319.
by yeldarb 6y ago
Yes, noted and agreed. In retrospect that came off a bit strong and I can see why it fanned the flames. Re the edit: we didn't name it, we just reported on it using the name that its creator chose.
320.
by yeldarb 6y ago
The original YOLO author publicly announced that he was no longer going to be working on computer vision models and has chimed in and said he has no problem with the name: https://twitter.com/pjreddie/status/127261
321.
by yeldarb 6y ago
Usually, it is not. There have been some attempts to combine image and text data into a hybrid model but I’m not sure how widespread it is. Ex: http://cbonnett.github.io/Insight.html
322.
by yeldarb 6y ago
Apologies if I appeared to be attacking anyone. That certainly wasn’t the intention.
323.
by yeldarb 6y ago
We’re not affiliated with any of the researchers. The many people taking issue with “v5” because it’s not by the same author as “v4” but not with “v4” even though it’s not the same author as “v3” are the “fanbases” I was referring to. FWIW,
324.
by yeldarb 6y ago
We’re also the top result when you google eg “How to train yolov4” and several of the top terms for training efficientdet. Hopefully we will be a great source of info on all computer vision models someday. Our mission is to make these thing
325.
by yeldarb 6y ago
Shameless plug: this is exactly why we built Roboflow. We’re trying to eliminate the one-off python scripts between labeling and training that everyone currently has to reinvent for themselves: https://roboflow.ai
326.
by yeldarb 6y ago
There's a huge gap between paying a couple hundred dollars for a pretty good model and investing hundreds of thousands (or millions) in hiring a team. Certainly if you're building something mission-critical, like self-driving cars
327.
by yeldarb 6y ago
Biggest surprise for me was that they can’t train on COCO. Would have thought their performance would be a major part of the marketing considering how prominent that benchmark is in the research community.
328.
by yeldarb 6y ago
Nope, we were able to use their UI to create our tests just like any other user would -- everything we needed was supported out of the box (minus one minor workaround I had to add to my code because I had some janky, non-deterministic IDs o
329.
by yeldarb 6y ago
We are using Reflect.run at Roboflow and it really is very slick. We have it testing our core flow on a schedule so we know if any of our 3rd party services or dependencies go down and/or if we introduce a regression. Our app's co
330.
by yeldarb 6y ago
In the business world we have trademarks for this. As models become brand names and franchises I wonder if academics will start trademarking them like many big open source projects do.

More ›