Tailoring to these areas mainly come from my personal focus areas, but at the same time, in the last 2 years the code and model sharing from research (both, university and companies/corporates) has leaped ahead dramatically in terms of providing code, models and sometimes datasets.
GitHub is so far the main source, but usually models and data is linked from many different resources due to large file sizes. And running/testing the approaches and models is often still quite cumbersome due to different versions and so on. Probably shared Google Colab notebooks is currently the least "painful" way to test out the methods.
That's the motivation to ask around and see if there maybe some other formats/platforms that would make such sharing even easier, thus reduce the "engineering" overhead in testing and help researchers and developers to focus more on the research part advancing the field even further :)
> And running/testing the approaches and models is often still quite cumbersome due to different versions and so on.
I definitely agree with this. I don’t know that I have any particular insight on infrastructure, but I think one of the things that makes huggingface and SMP [0] successful is a well defined common model and data API.
It can be hard to clone a random research repo and train models on custom datasets because the model API is slightly different, or expects the data in a different format. I think this is a thing in graph neural networks right now, where different libraries use different data formats, and some groups roll their own
Technologies: Robotics, Deep Learning, CNNs, Computer Vision, OpenCV, ROS, C++, Python, Hands on experience with UR robots, Kuka and Franka Panda, getting into Data Science, some web development also. Currently finishing PhD in Robotics.
Worked on a number of research and commercial products, all with practical applications. Work experience in Lithuania, UK, Switzerland, Australia and Austria, so absolutely no problem to work in diverse and dynamic teams. Education completed:
Interesting idea, thanks! gonna think on how rfid or similar tagging can help to avoid adding complex CV algorithms where much easier alternatives could be offered. There are some robots moving around trolleys with items already there, so staff and patients are getting used to see machines sharing same corridors
New rfid applications would be great and by means pursue them! The reason I think this is a great problem for CV is that right now, only a nurse would be able to tell if there is a puddle on the floor from a leaking IV, patient's puke, overflown catheter bag, etc. Also is there air in a line, does the label on the drug match the prescription in EMR? These are common sense, cheap to fix / prevent problems that I'm not sure rfid could solve. RFID would be great for matching drug labeled with rfid with chip in patient's armband. One could think of lots of implementations for that sort of thing!
I guess it really comes down to what sort of focus you want your project to take, and what idea you think your PI will be interested in considering. Good luck!
Interesting idea. However, I believe doctor correction can be a very delicate matter as professionals are still not happy of machines pretending to "know better". However, we did discuss some surgeon assisting robots, like holding various tools or ultrasound scanner in place and adjusting for patient's body movements
Really cool app, thanks! I've been using it the whole day yesterday and performance was really good (did get few false positives though..). I was pleasantly surprised when it recognised next song gestures in really dark conditions, hand was mainly lit up just from the macbook screen itself.
It would be really cool to have a gesture for answering a Skype call (maybe even making a call, to pop up online contact list). Telephone hand gesture should do the trick.
Keep up the good work! I can see this going beyond the music control.
Hot Air Balloon tourism in Vilnius is gaining popularity due to relatively low price and the unique opportunity to fly over the city. Only two more cities in Europe allow Hot Air Balloon flights over the city - Helsinki and Stockholm. It’s a very rare occasion that flights are allowed over the the international airport, not only Europe, but probably anywhere in the world.
VNO approach Air Traffic Control (ATC) has special arrangements with balloon pilots allowing them to fly in one flock, they have to take off within 10 minute interval and all the communication is done with the leader of the group. Once the balloon is right over the airport territory, it has to ascent to 1km height in order to allow planes to go under, while in the close proximity it’s opposite, they go down allowing planes to fly over.
In general, such flights are possible only because of good relationship between Hot Air Balloon pilots and ATC who have full control of the situation, so it happens even when some planes approaching the airport. Every situation is handled individually.
Tailoring to these areas mainly come from my personal focus areas, but at the same time, in the last 2 years the code and model sharing from research (both, university and companies/corporates) has leaped ahead dramatically in terms of providing code, models and sometimes datasets.
GitHub is so far the main source, but usually models and data is linked from many different resources due to large file sizes. And running/testing the approaches and models is often still quite cumbersome due to different versions and so on. Probably shared Google Colab notebooks is currently the least "painful" way to test out the methods.
That's the motivation to ask around and see if there maybe some other formats/platforms that would make such sharing even easier, thus reduce the "engineering" overhead in testing and help researchers and developers to focus more on the research part advancing the field even further :)