Tutorial Overview - Custom Semantic Segmentation p.1
ฝัง
- เผยแพร่เมื่อ 4 ต.ค. 2024
- The first video in a semantic segmentation tutorial series. Check out the code here: github.com/set.... This is meant to provide resources for programming semantic segmentation as well, so watch specific videos to meet your needs.
![[UNCUT]"ตุลามรณะ"ย้อนฟังคำทำนาย หมอดูทุกสาย ดวงเมืองร้อน (หมอปลาย ฟองสนาน อ.โอเล่)I คนดังนั่งเคลียร์](http://i.ytimg.com/vi/lE5az5VvT9M/mqdefault.jpg)
I wish I could explain how thankful I am that you made this series. You’re a life saver and your explanation made it so much simpler
Awesome! I'm not expecting many people to see this series and the models wouldn't really be used in production these days but I'm glad some people learned from it.
at 4:40 when you pulled a list of what the entire series is about you gained my undivided attention
This is incredibly well-organized and well-explained. Thanks so much for providing it!
🤘🤘🤘Gonna follow the series and implement it whilst watching...thank you for the series ❤️
Glad to see your new series !!
Thanks! I think this series is much better in regards to how I formatted everything than the last one. Plus I provided all the code this time.
@@seth8141 if you would like, i have the previous series in jupyter notebooks
Nah it's fine. It's not that good of a series anyway. I didn't even do the audio processing on the gpu. If I ever do audio stuff in the future I will just improve everything.
Great Job Seth...Thanks and keep doing that. I See you from Spain!!!
I also did a review in github and it is amazing, very well explained !!!
Coffee check, notebook check, sound check....lets set go :)
Thank you!
Very well, a simple question? From this tutorial can I genterate a .pt model? I need use my model on Android..
Awesome video thanks for sharing.. can we measure the segments area? Like percentage of the circle, square...etc
Yeah depending on what you need you can sometimes achieve this without all the semantic segmentation. See: docs.opencv.org/3.1.0/dd/d49/tutorial_py_contour_features.html. Obviously you would like to identify the shapes though.
Just realize the area is also a function of depth (from camera to scene), so you can do some calibrating to account for that, but it is just a single camera.
Do you have tutorial for orbit semantic segmentation?
Hi Seth! I would like to get a summary generated for say this very TH-cam video, how to achieve it? I tried many translation tools but they weren’t really accurate, I see subtitles coming up for every video frame, in a similar way can I not get brief description of the video based on the stress the speaker puts on some sentences, number of times he says something or some sort of automation.
If at all this is a ML/DL problem, what would that be. NLP or CV and how to solve it? Your response is greatly appropriated.
It's more of just a deep learning problem. On a lower level you could start with image captioning. (Neural talk for example) idk you'd have to check the latest papers. Sounds difficult for a whole video.
@@seth8141 I am sorry for my ignorance but does image captioning apply in this case because we're talking about a recorded video which has both video and voice. Now keeping the video aside can we not just translate the voice into text. I wonder if neuraltalk can help as it again works to describe images with sentences, but not video. I'm confused! Can you advise please.
@@ramkotha4726 I've only ever done it with describing images. For video you would need something more advanced since this information occurs many frames per second in addition to adding in text transcribed from audio. But just take a frame every second would probably be a good place to start. You should review some publications to see what is out there.
@@seth8141 Thanks very much, that helps indeed. Based on your description of the problem I'm assuming this is even a harder problem than what I think. I'll look for some papers on this, could you also share any link or information if you come across. Really appreciate it.
Just sent you an Email! thanks!
sucks... you spoke of nothing.