Тёмный
Karol Majek
Karol Majek
Karol Majek
Подписаться
Karol Majek

karolmajek.pl
YOLOv7 e6 1280px
25:31
2 года назад
YOLOv7 d6 1280px
25:31
2 года назад
YOLOv7 e6e 1280px
25:31
2 года назад
YOLOv7 640px
25:31
2 года назад
YOLOv7 w6 1280px
25:31
2 года назад
YOLOv7 640px
25:31
2 года назад
SegFormer B5 1024px @ 1024px inference
30:37
2 года назад
SegFormer B5 1024px @ 2048px inference
30:37
2 года назад
OpenVINO SSD MobileNet V2
30:37
2 года назад
OpenVINO SSDLite Mobilenet V2
30:37
2 года назад
YOLOR P6 2048x1152
25:31
3 года назад
YOLOR E6 2048x1152
25:31
3 года назад
YOLOR ssss DWT 2048x1152
25:31
3 года назад
YOLOR ssss s2d
25:31
3 года назад
YOLOP @ 1280x736 on Titan RTX
30:37
3 года назад
YOLOP @ 640
30:37
3 года назад
YOLOR W6 2048x1152
25:31
3 года назад
SOTR R101
30:37
3 года назад
SOTR R101 DCN
30:37
3 года назад
YOLOR D6 2048x1152
25:31
3 года назад
NanoDet-m 416 - RTX2070 laptop FP16
30:37
3 года назад
SETR PUP 512x512 160k ADE20k batch 16
30:37
3 года назад
TensorRT_demos YOLOv4 CSP 512
30:37
3 года назад
SETR PUP 768x768 80k Cityscapes batch 8
30:37
3 года назад
TensorRT_demos Tiny YOLOv4 416
30:37
3 года назад
Комментарии
@SeaUsername
@SeaUsername 25 дней назад
is the original video still available for download ?
@nazilamalekzadah4039
@nazilamalekzadah4039 25 дней назад
I need a dataset for car accident , like 10,000 pictures . Do you know where I can find it ?, cuz I searched a lot but coudn't find one
@romanemul1
@romanemul1 2 месяца назад
Good work. i tried to train with CenterNet and Faster RCNN but the results were nowhere good as with SSD.
@galalatif
@galalatif 3 месяца назад
just for validation, How many cars in the video?
@rickykim4193
@rickykim4193 4 месяца назад
Hi, can I also access linear velocity of the camera along the x, y, z axis? If so, which topic can I access them?
@experttreeremoval2667
@experttreeremoval2667 4 месяца назад
What type of dashcam did you use to record it? What type of quality do you need?
@jorgedominguez8440
@jorgedominguez8440 6 месяцев назад
Hello friend, I want to read the size of a tire on the sidewall, which you recommend I use. excellent videos, thank you.
@yippeedigital3906
@yippeedigital3906 7 месяцев назад
I'm reaching out because I'm interested in tracking the number of vehicles and pedestrians. I'm curious if you have a solution that could be applied to other sites as well. Additionally, I'm concerned about privacy and wanted to inquire if any personal data is collected, and if so, whether it can be eliminated from the process. Looking forward to hearing from you.
@yippeedigital3906
@yippeedigital3906 7 месяцев назад
I'd be interested in talking to you about this for my company. Do you have a workable solution, and is it easily applied on any site?
@anren7445
@anren7445 8 месяцев назад
thank you for this
@ahf733
@ahf733 8 месяцев назад
Thank you very much, your explanations were clear and I could run it 🎉
@websiteflixwebagency3689
@websiteflixwebagency3689 8 месяцев назад
Is there a way to identify vehicle make, model and color?
@崔维帅-w1l
@崔维帅-w1l 10 месяцев назад
How can you use your own dataset to train the ESPNet model?
@崔维帅-w1l
@崔维帅-w1l 10 месяцев назад
你好
@Artificial_Intelligence_PIAIC
Haar cascade object detection 14-AUG-2023 ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-raRTB_jqzMw.html
@Dr_Techo
@Dr_Techo Год назад
Greetings. Can I have such a video that runss for 24 hours or more? Thank you.
@patagonia4kvideodrone91
@patagonia4kvideodrone91 Год назад
what is see in the sky in the min 6? from right to left
@yacinerouizi844
@yacinerouizi844 Год назад
thank you
@bikcrum
@bikcrum Год назад
I am trying to mount them in bipedal robot Cassie (ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-a_YGPbWJO5g.html) but since the motion is too jerky, the camera is losing its tracking. Do you recommend any method that can help stabilization?
@waynehu7865
@waynehu7865 Год назад
Hello, I will use the intel d435i camera lens to make a special topic. I have encountered some problems. Can I ask you for advice?
@ED.UU2006
@ED.UU2006 Год назад
Vim pelo chat gpt ele mandou um link 😅
@anhtran1989
@anhtran1989 2 месяца назад
🎉
@yuqianglin4514
@yuqianglin4514 Год назад
Has anyone tried to use this video for number plate detection? Is it available?
@FaceSenseAustralia
@FaceSenseAustralia Год назад
I don't think this video is very useful for number plate detection. however, I have made a traffic control management system by using similar type of video.
@segri544
@segri544 Год назад
thanks you. Any idea about height of the cam and angle between horizantal?
@dreadhawk123
@dreadhawk123 Год назад
So you can get OpenVSLAM working but you can't use screen-recording software?🤣
@codewithguillaume
@codewithguillaume Год назад
Hello Karol, thanks for this video. Is it possible to apply it to a live video? Like a camera? On a Raspberry Pi 4? Dumb question maybe. Thanks !
@codewithguillaume
@codewithguillaume Год назад
My friend that's so interesting. Thanks !
@piotr780
@piotr780 Год назад
Hi, do you know any good description of FCOS except the paper itself ? (I'm reading it now, and it's quite good)
@rgruberski
@rgruberski Год назад
Hi! What camera did you use to record this video?
@KarolMajek
@KarolMajek Год назад
It was long time ago. I think it was Samsung S7
@svetoslavarsov7327
@svetoslavarsov7327 2 года назад
Impressive! Excellent job. But how far are computer vision with AI from human intelligence? At time 0.38 in the video there is a mirrored car in shop's window and the program mark as a car. May in the future AI need a real 3d computer vision (with depth). How will calculate all this 3d vision in real time and how much computer power is need for 3d model learning? Maybe quantium computing is a right key.
@ShopperPlug
@ShopperPlug 2 года назад
What is the "VOC" sample?
@햇고구마-v3e
@햇고구마-v3e 2 года назад
I want to run Your Lanenet, but can't do import Lanenet.
@KarolMajek
@KarolMajek 2 года назад
That's not mine code. Did you installed it? Are you in proper directory?
@fackdog
@fackdog 2 года назад
*Top!*
@PusikPark
@PusikPark 2 года назад
What kind of 360 camera did you use?
@twiekfkwniejwjrj7867
@twiekfkwniejwjrj7867 2 года назад
what is the mask?
@piyush9555
@piyush9555 2 года назад
how precisely accurate computers have become, we're in a real trouble now
@zhiliu6
@zhiliu6 2 года назад
what about the speed?
@TheChenchen
@TheChenchen 2 года назад
If you only read the description.
@GolemTH9
@GolemTH9 2 года назад
would you please make a simple method to coding tracking with yolov3 tiny
@tiro0oO5
@tiro0oO5 2 года назад
Where are the 3D BB
@KarolMajek
@KarolMajek 2 года назад
In v2? This is the official demo
@tiro0oO5
@tiro0oO5 2 года назад
@@KarolMajek was just kidding.
@TheRubiksrocks
@TheRubiksrocks 2 года назад
Re-upload?..
@netnomadaorta8711
@netnomadaorta8711 2 года назад
the whole image is transformed into 640px X 640px??
@pmioduszewski
@pmioduszewski 2 года назад
Nice! 😃
@23y75
@23y75 2 года назад
the streetlights being detected as umrellas, airplanes and kites were funny
@rodrigocardoso7846
@rodrigocardoso7846 2 года назад
Are you using NMS or other method with yolo?
@DanFrederiksen
@DanFrederiksen 2 года назад
and this is a stateless classifier? no frame to frame memory
@DanFrederiksen
@DanFrederiksen 2 года назад
it's getting pretty solid. although stop sign detection is pretty loose. how many params?
@23y75
@23y75 2 года назад
the prohibition signs in poland (and other EU) having a red outline is maybe throwing it off
@KarolMajek
@KarolMajek 2 года назад
Out of distribution (only stop sign in the dataset)
@laythabdulkareem1887
@laythabdulkareem1887 2 года назад
You are amazing!
@dlcarlosrojas1036
@dlcarlosrojas1036 2 года назад
Hello very nice channel. I have a question... How did you make cars and people have the same color? I have read all the documentation, and it seems that modifications must be made in visualizer.py I hope you can share your experience with me. THANK YOU
@DanFrederiksen
@DanFrederiksen 2 года назад
Looks quite robust on vehicles, particularly for the size. How many parameters does it have? hard to find anywhere. They should always state model size. It's quite impressive for how many categories it can detect at such size and speed although it failed quite hard on a stop sign. I assume it runs on monochrome? no color data.