YOLO-V4: CSPDARKNET, SPP, FPN, PANET, SAM || YOLO OBJECT DETECTION SERIES

This video is about Yolo object detection family. This is about YoloV4 which is the most popular and widely used object detector in the industry. YoloV4 has the highest usage by industry for commercial purposes because of its optimal speed and accuracy. In this video, we discussed about Backbone CSPDarknet-53, SPP, FPN, PANT and SAM modules. These are all parts of Bag of Specials in YoloV4.
YOLO Playlist:
• YOLO OBJECT DETECTION ...
Neural Networks From Scratch Playlist:
• Neural Networks From S...
Link to Papers:
YoloV4: arxiv.org/pdf/2004.10934.pdf
DenseNet: arxiv.org/pdf/1608.06993.pdf
CSPNet: arxiv.org/pdf/1911.11929.pdf
FPN: arxiv.org/pdf/1612.03144.pdf
PANET: arxiv.org/pdf/1803.01534.pdf
Mask-RCNN: arxiv.org/pdf/1703.06870.pdf
SCRDet: arxiv.org/pdf/1811.07126.pdf
SAM: arxiv.org/pdf/1807.06521.pdf
SPP: arxiv.org/pdf/1406.4729.pdf
Chapters:
00:00 Introduction
00:39 Topics Covered in this video
00:53 YoloV4 Backbone
02:15 Dense Block
03:49 DenseNet Architecture
06:51 CSPNet Intuition
08:00 CSP + DenseNet
10:15 CSP + DarkNet53 & Mish
11:24 Need of Neck Module (Intuition)
18:15 FPN Intuition
21:30 PAN Need & Intuition
27:34 Adaptive Feature Pooling Intuition
35:25 Adaptive Feature Pooling Explanation
38:24 Modified PAN in YoloV4
39:40 Spatial Pyramid Pooling
44:08 Attention Mechanism Intuition
46:40 Spatial Attention Module (SAM)
49:26 Modified SAM in YoloV4
50:42 Conclusion
#yolo #yoloobjectdetection #objectdetection #yolov4 #yolov5 #yolov3 #yolov7 #computervision #imageclassification

[@prashantbhardwaj4831]

One of the best I have come across so far, Could you please continue the series.

[@kundankumarmandal6804]

Your lecture is one of the best i have seen so far....very informative,clear,to the point and interesting as well. Really awesome keep doing the good work...

[@Teo-uw7mh]

Sir, im loving the series, please continue the great work

[@MLForNerds]

Thanks, will do!

[@priyamanikandan2473]

Sir kindly continue the series till YOLOV8. Your series is amazing and we are awaiting for you to complete the series

[@zeeshansheikh8163]

Really Enjoyed learning these Yolo Version -- Still waiting for remaining versions

[@MLForNerds]

Will continue in sometime.

[@AndreiChegurovRobotics]

what a great material - so well structured. thx for your efforts - beautiful!!!

[@nikitamohod7971]

Thank you for providing this video. kindly provide such type of video on YOLOv5 , YOLOv6 and YOLOv7 also.

[@MLForNerds]

Will keep posting on those one by one

[@taojiang2735]

@@MLForNerds looking forwards to that, when will YOLOV5 be released

[@giabao2602]

thanks for giving us perfect video, love you bro

[@yashbohra2593]

Really informative and precised content.

[@tanmaythaker2905]

Amazing Session!

[@MLForNerds]

Glad you enjoyed it

[@sam-vv6gl]

Thanks for the great explanation

[@ansariarbaz3374]

Please continue the series ASAP sir

[@edwardsnowden8749]

Hi, could you please post a video about polygon annotation for objects, and which YOLO version supports that, and how to implement that?

[@sainatha1785]

sir please make the content for remaining all versions and object detection models, videos are very captivating sir

[@gareven]

Does sam-mish activation function have a formula ? or it's same with mish

[@madhav5166]

Just like sir said the higher level feature map or smalle feature map are mainly for bigger object and lower level feature map or bigger feature map are for smaller object,due to this various scale are neded for differenet feature map , can anyone tell me in simple words what is 'scale' here

[@user-uf3md5ub5j]

Thanks a ton for the nice videos, will be any following videos in the series?

[@MLForNerds]

Yes, I will be releasing videos on all YOLO versions.

[@user-uf3md5ub5j]

@@MLForNerds 🙏❤

[@ms-dq6hk]

Hello! thank you for the video. I think there are some innacurate concepts, for instance, there is not region of interests here because yolo is a one-stage detector. So the ROIs you are talking about, where they came from?

[@MLForNerds]

Hi thanks for the feedback. Can you tell me the timestamp you are talking about?

[@bintangrjp2183]

Thankyou for your explanation. I’m currently using yolo v4 for my thesis reserach and do you have some recommendation how many batch and subdivision should i use for training? becuase i keep getting low mAP after training using yolo v4 which is 79%, and i have 450 data for training, is there any chance that i can go higher than that? and how? thank you

[@lotze_cinema]

you actually need more data bc it says you need at least 1500-2000 objects per class

[@bugdary]

Great video. Thanks ! Can you provide a link to part 3 ?

[@MLForNerds]

I'm working on it. Will be posting next week mostly.

[@neeru1196]

Eagerly waiting :)
@@MLForNerds

[@hamzahra5063]

could you please share with us the pdf or ppt version of this presentation

[@computervisionwithandy8181]

Hello sir, can you make videos on latest YOLO version such as V8 and V7, thank you!!

[@MLForNerds]

Sure, will try to upload.

[@arpitsahni422]

hi great work! can you share slides

[@kaboomofficial3473]

hii i have a doubt is there a way i can contact you i have few doubts regarding neural networks that i hope you can solve as i saw your videos and they were pretty awesome

[@MLForNerds]

Yes, sure. I have started discord server recently. I’m sharing the link and you can join there. discord.gg/nfZCszw7Rw

[@kaboomofficial3473]

@@MLForNerds okie

[@ramdharavath7542]

Please upload yolov7

[@Yash_-cf6he]

bro YOLOv5 please

[@ROHANKUMAR_]

yes,bro

[@ROHANKUMAR_]

please upload yolov5

[@mallaswetha5629]

please upload yolov5

[@MLForNerds]

I will be posting in the coming month.