YOLOv1 from Scratch

Поделиться
HTML-код
  • Опубликовано: 7 янв 2025

Комментарии • 306

  • @AladdinPersson
    @AladdinPersson  4 года назад +58

    Here's the outline for the video:
    0:00 - Introduction
    0:24 - Understanding YOLO
    08:25 - Architecture and Implementation
    32:00 - Loss Function and Implementation
    58:53 - Dataset and Implementation
    1:17:50 - Training setup & evaluation
    1:40:58 - Thoughts and ending

  • @PaAGadirajuSanjayVarma
    @PaAGadirajuSanjayVarma 4 года назад +92

    Plz give this man a noble proze

    • @deeps-n5y
      @deeps-n5y 3 года назад

      *Nobel

    • @iiVEVO
      @iiVEVO 3 года назад +4

      A noble nobel prize*

  • @vijayabhaskar-j
    @vijayabhaskar-j 4 года назад +96

    This series was super helpful, can you please continue this by making one for Yolo v3, v4, SSD, and RetinaNet? That will make this content more unique because none of the channels that explains all these architectures and your explanations are great!

    • @jertdw3646
      @jertdw3646 2 года назад

      I'm confused on how i'm supposed to load the images up for training. Did you get that part?

    • @Glitch40417
      @Glitch40417 Год назад

      ​​@@jertdw3646on't know if you got it or not, actually there's a train.csv file.
      Instead of 8examples.csv or 100examples.csv we can use that file.

  • @asiskumarroy4470
    @asiskumarroy4470 4 года назад +14

    I dont know how do I express my gratitude to you.Thanks a lot brother.

  • @MohamedAli-dk6cb
    @MohamedAli-dk6cb 2 года назад +14

    One of the greatest deep learning videos I have ever seen online. You are amazing Aladdin, please keep going with the same style. The connections you make between the theory and the implementation is beyond PhD level. Wish I can give you more than one like.

  • @_adi_1900
    @_adi_1900 4 года назад +9

    This channels going to blow up now. Great stuff!

  • @thetensordude
    @thetensordude 4 года назад +55

    Most underrated channel!!!

    • @vanglequy7844
      @vanglequy7844 3 года назад

      Let's look at it upside down then!

  • @keshavaggarwal5835
    @keshavaggarwal5835 4 года назад +3

    Best Channel ever. Cleared all doubts about YOLO. I was able to implement this in tensorflow by following your guide with ease. Thanks a lot bro.

    • @AladdinPersson
      @AladdinPersson  4 года назад +1

      Awesome to hear it! Leave a link to Github and people could use that if they are also doing it for TF?:)

    • @Skybender153
      @Skybender153 3 года назад +1

      Link for the tensorflow repo would be appreciated Keshav

  • @caidexiao9839
    @caidexiao9839 2 года назад +3

    Thanks a lot for you kindness to provide the yolov1 video. By the end of the video, you got mAP close to 1.0 with only 8 training images. I guess you used weights of a well trained model. With more than 10,000 images and more than 20 hours on Kaggle 's free GPU, my mAP was about 0.7, but my validation mAP was less than 0.2. Nobody mentioned the over fitting issue of yolo v1 model training.

    • @satvik4225
      @satvik4225 7 месяцев назад +2

      mine is coming 0.0 always

    • @TornadoFilms_
      @TornadoFilms_ 20 дней назад

      @@satvik4225 yeea why is that , did u got that fixed

  • @Anonymous-nz8wd
    @Anonymous-nz8wd 4 года назад +4

    GOD DAMN! I was searching for this for a really long time but you did it, bro. Fantastic.

  • @haldiramsharma4601
    @haldiramsharma4601 4 года назад +8

    Best channel ever!! All because of you, I learned to implement everything from scatch!! Thank you very much

  • @eminemhc5763
    @eminemhc5763 4 года назад +5

    Only 3.5K subscribers ??? One of the most underrated channel in RUclips
    Keep posting quality video like this bro , soon you will reach 100K+ subs , congrats in advance
    Thanks for the quality content :)

  • @nguyenthehoang9148
    @nguyenthehoang9148 Год назад +1

    By far, your series is one of the best content about computer vision on RUclips. It's very helpful when people explain how things work under the hood, like the very well-known courses by Andrew Ng. If you make a paid course for this kind of content, I'll definitely buy it.

  • @thanhquocbaonguyen8379
    @thanhquocbaonguyen8379 3 года назад +7

    massively thank you for implementing this in pytorch and explain every bits in detail. it was really helpful for my university project. i have watched your tutorials at least 3 times. thank you!

    • @abireo2285
      @abireo2285 2 года назад

      PhDs are 100% learning how to code here :)

  • @_nttai
    @_nttai 4 года назад +3

    I was lost somewhere in the loss but still watch the whole thing. Great video. Thank you

  • @rampanda2361
    @rampanda2361 3 года назад +1

    The savior, Been looking at codes of other people for few days, Could not understand it better as those were codes only with no explanation what so ever. Thank you very much.

  • @sangrammishra4396
    @sangrammishra4396 2 года назад +1

    I love the way he explained and always maimtain simplicity in explaining the code, thanks aladdin

  • @crazynandu
    @crazynandu 4 года назад +14

    Great Video as usual . Looking forward to see RCNNs (mask , faster , fast , ..) from scratch from you !! Similar to Transformers you did, you can do one from scratch and other using the torchvision's implementation .Kudos !!

  • @Тима-щ2ю
    @Тима-щ2ю 8 месяцев назад

    What an amount of work! I don't often see people in the internet that are so dedicated to deep learning!

  • @sachavanweeren9578
    @sachavanweeren9578 2 года назад +2

    I can imagine this video took a lot of time to prepare, the result is great and super helpful. Thank you very much. Respect!

  • @abireo2285
    @abireo2285 2 года назад

    This is the best deep learning coding video I have ever seen.

  • @krzysztofmajchrzak1881
    @krzysztofmajchrzak1881 4 года назад +1

    I want to thank so much! It is literally a live saver for me! Your channel is underrated!

  • @WiktorJurek
    @WiktorJurek 3 года назад +3

    This is insanely valuable. Thank you very much, dude.

  • @张子诚-z3b
    @张子诚-z3b 3 года назад

    I'm a beginner of object detection, You videos help me a lot. I really like your style of code.

  • @nikolayandcards
    @nikolayandcards 4 года назад +3

    So glad I came across your channel (Props to Python Engineer). Very valuable content. Thanks for sharing and you have gained a new loyal subscriber/fan lol.

  • @ai4popugai
    @ai4popugai Год назад

    The most clear explanation that I have ever found, thank you!!

  • @ИльяЯгупов-н4я
    @ИльяЯгупов-н4я Год назад

    Thank you so much for this video, it's so helpful! Especially the concept in first 9 minutes. I read a lot of sources, but here it's the only place where it is clearly explauned. And more precisely the part where we are looking for a cell with midpoint of bounding box! Thank you so much for a great Explanation!

  • @pphuangyi
    @pphuangyi Год назад

    Thanks!

  • @shantambajpai8064
    @shantambajpai8064 4 года назад +2

    Dude, this is AMAZING !

  • @vil9386
    @vil9386 14 дней назад

    Absolutely awesome. Paper to python code is such a valuable teaching input for aspiring AI/ML engineers.

  • @정래혁-c8y
    @정래혁-c8y 3 года назад +2

    This video was so helpful. Thank you!

  • @majtales
    @majtales 4 года назад +1

    @27:05 why flatten again? Isn't it already flattened in the forward method of the class?
    Also, do we really need to flatten? @51:22 The MSELoss documentation says it sums over all dimensions by default. Also how did you work around that division by zero?@1:33:15

  • @poojanpanchal3721
    @poojanpanchal3721 4 года назад

    Great Video!! never seen anyone implementing a complete YOLO algorithm from scratch.

  • @vishalm2338
    @vishalm2338 4 года назад

    Thanks a ton Aladdin for making this video. I truly loved it. Also, Would like to see Retinanet implementation . It would be really fun to watch too. Kudos to you!!

  • @zachhua7704
    @zachhua7704 3 года назад +4

    Hi Aladdin, thanks for the great tutorial. I got a question at 1:13:09, in the paper, authors say the width and height of each bounding box are relative to the whole image, while you say they are relative to the cell. Is that a mistake?

  • @TheDroidMate
    @TheDroidMate Год назад

    Amazing video series, thanks! Extra kudos for the OS you're using 💜

  • @ignaciofalchini8264
    @ignaciofalchini8264 3 года назад

    you are awesome bro, really nice job, best YOLOv1 video in existence, thanks a lot

  • @francomozo6096
    @francomozo6096 4 года назад

    Thank you man!!!! Great video! Gave me a really good understanding on Yolo, will subscribe

  • @sumitbali9194
    @sumitbali9194 3 года назад

    Your videos are a great help to data science beginners. Keep up the good work 👍

  • @haideralishuvo4781
    @haideralishuvo4781 4 года назад

    FInally , Most waited video , Will have a look asap

  • @1chimaruGin0_0
    @1chimaruGin0_0 4 года назад +2

    Great work as always!
    This video help me a lot to understand my confusion about yolo loss.
    Could you do some video on Anchors and Focal loss?

    • @AladdinPersson
      @AladdinPersson  4 года назад +2

      I'll revisit object detection at some point and try to implement more state of the art architectures and will look into it :)

  • @santoshwaddi6201
    @santoshwaddi6201 3 года назад

    Very nicely explained in detail.... Great work

  • @bradleyadjileye1202
    @bradleyadjileye1202 Год назад

    Absolutely wonderful, thank you very much for such a fantastic job !

  • @Old_SDC
    @Old_SDC Год назад

    Will be back, just need a quick break 35:30
    Downloading 59:42

  • @pixarlyVII
    @pixarlyVII 3 года назад +1

    I have a question. At 39:41 you, from utils, import intersection_over_union. I thought that dataset.py, loss.py, ..., utils.py where empty python files. Why did you imported a function from utils.py if in the tutorial we dont code anything in this file?
    I've followed the tutorial and Im stucked at 59:50 bc my code cant import name "intersection_over_union" from "utils".

    • @pixarlyVII
      @pixarlyVII 3 года назад

      Nada, soy gilipollas. Me he copiado el archivo utils.py de lo que has subido a GitHub y ya va.
      It would be interesting to code that part (utils.py) too in the tutorial.

  • @NamNguyen-fn5td
    @NamNguyen-fn5td 3 года назад +1

    Hi. I have question at 1:12:29. Why "x_cell, y_cell = self.S * x - j, self.S * y - i" minus j and i ? What does this mean?

    • @NamNguyen-fn5td
      @NamNguyen-fn5td 3 года назад

      at 50:27 if you not flatten box_predictions and box_target in MSEloss, it is the same result as flatten

  • @changliu3367
    @changliu3367 3 года назад

    Awesome video. Pretty helpful! Thanks a lot.

  • @ilikeBrothers
    @ilikeBrothers 3 года назад +1

    Просто топчик! Огромное спасибо за столь подробное разъяснение ещё и с кодом.

  • @patloeber
    @patloeber 4 года назад

    Amazing effort!

  • @mizhou1409
    @mizhou1409 3 года назад

    Great job, very helpful for a new beginner.

  • @SamtapesGamer
    @SamtapesGamer Год назад

    Amazing!! Thank you very much for all these lessons! It would help me a lot if you could make videos implementing Kalman Filter and DeepSort from scratch, for object tracking

  • @jitmanewtyagi565
    @jitmanewtyagi565 3 года назад +1

    Broooooo, thanks for this man.

  • @vamsibalijepally3431
    @vamsibalijepally3431 4 года назад +1

    def test(S=7, B=2,C=20):
    model = Yolov1(in_channels=3,split_size=S,num_boxes = B,num_classes=C)
    x = torch.randn((2,3,448,448))
    print(model(x).shape)
    this will throw help if got same error like me
    __init__() missing 1 required positional argument: 'kernel_size'

    • @pranavkushare6788
      @pranavkushare6788 4 года назад

      Yeah i'm getting the same error.
      Have you found any solution and reason ?

    • @chinmay996
      @chinmay996 3 года назад

      @@pranavkushare6788 if you still have not solved the problem, check your parameters in CNNBlock inside _create_conv_layers method.

  • @vikramsandu6054
    @vikramsandu6054 3 года назад

    Your name is Aladdin but you are a genie to us. Thanks for this video.

  • @sb-tq3xw
    @sb-tq3xw 4 года назад

    Amazing Work!!

  • @soorkie
    @soorkie 4 года назад +7

    Hi, can you do a similar one with Graph Convolutional Networks? Your videos are very usefull ❤️

  • @PaAGadirajuSanjayVarma
    @PaAGadirajuSanjayVarma 4 года назад

    I am glad I found your channel

  • @buat_simple_saja
    @buat_simple_saja 2 года назад

    Thank you man, your video help me a lot

  • @anierrn6935
    @anierrn6935 3 года назад

    35:35 explanation about square roots for w,h

  • @Epistemophilos
    @Epistemophilos 2 года назад +6

    Is there a mistake in the network diagram in the paper? Surely the 64 7x7 filters in the first layer result in 64 channels, not 192? What am I missing? If it is a mistake (seems highly unlikely), then the question is if there are really 192 filters, or 64.

    • @chocorramo99
      @chocorramo99 5 месяцев назад +1

      64 kernels and there are 3 channels, 192 resulting channels. lol kinda late.

    • @Epistemophilos
      @Epistemophilos 5 месяцев назад +2

      @@chocorramo99 Linear algebra is timeless! Thanks :D

  • @nova2577
    @nova2577 4 года назад

    Appreciate your effort!!

  • @GursewakSinghDhiman
    @GursewakSinghDhiman 3 года назад

    You are doing an amazing job. Thanks alot

  • @jaylenzhang4198
    @jaylenzhang4198 Год назад

    My understanding of this λ_noob-associated loss function is that it is used to penalize false negatives. This λ_noob-associated loss function includes all grid cells that do not contain any objects but have confidence scores larger than 0. Since there will be a lot of these false negatives, the author adds the coefficient λ_noob to lower their ratio in the overall loss function.

  • @proxyme3628
    @proxyme3628 Год назад

    Regarding the loss for the confidence (FOR OBJECT LOSS part in loss.py), the label Ci should be IoU? In the code, it is (torch.flatten(exists_box * target[..., 20:21]), but because exists_box is target[..., 20:21], it is just a square of target[..., 20:21]? The original v1 paper said "Formally
    we define confidence as Pr(Object)   IOUtruth
    pred . If no
    object exists in that cell, the confidence scores should be
    zero. Otherwise we want the confidence score to equal the
    intersection over union (IOU) between the predicted box
    and the ground truth", which suggests the Ci_hat is to be calculated from IoU.

  • @leochang3915
    @leochang3915 4 года назад

    Thank you , you really help me a lot!

  • @adarshsingh936
    @adarshsingh936 3 года назад +2

    Can someone explain the use of unsqeeze(3) at 43:55

  • @ZXCOLA-z7s
    @ZXCOLA-z7s 2 года назад

    That’s totally awesome!

  • @qichongxia2110
    @qichongxia2110 11 месяцев назад

    very helpful! thank you !

  • @krishnasumanthmannala984
    @krishnasumanthmannala984 4 года назад

    At 03:42 the width and height of an object are relative to the image I think wrt YOLO 1.

  • @canyi9103
    @canyi9103 Год назад

    4:24, In paper the width and height are predicted relative to the whole image. they can not be larger than 1, but in your video, you said it can larger than 1. It seems not right

  • @jeroenritmeester73
    @jeroenritmeester73 3 года назад +3

    How does the very first layer of the DarkNet with out_channels = 64 produce 192 feature maps? I understand that 3*64 = 192 but I don't really see how that applies.
    Similarly, the second step has a convlution of 3x3x192, but there are 256 feature maps afterwards.

    • @DanielPietsch-o6r
      @DanielPietsch-o6r Год назад

      I am also confused about that part. In my understanding it should be 7x7x3 and then 192 total kernels, right?

  • @heriun7268
    @heriun7268 3 года назад

    4:00 I think you are wrong. w,h is realative to the whole image. check paper Section 2.Unified Detection - 4th paragraph

  • @venkateshvaddadi271
    @venkateshvaddadi271 3 года назад

    great job brother
    you are really awesome

  • @vijayabhaskar-j
    @vijayabhaskar-j 4 года назад

    at 42:13 shouldn't that be [...,25:29] not [...,26:30] as the first iout_b1 covers 21,22,23,24 and the second should cover 25,26,27,28? or 25th is the confidence score and 26,27,28,29 are the second bounding boxes?

    • @AladdinPersson
      @AladdinPersson  4 года назад +2

      Yes you're correct, 25th is for the confidence score for the second bbox and 26:30 (remember it's non-including the 30th index) so I think what is shown is correct

  • @hetalivekariya7415
    @hetalivekariya7415 2 года назад

    Why I did not come across your channel before!!. But anyways I am glad I found your channel. Thank you.

  • @bhavyashah8674
    @bhavyashah8674 2 года назад +1

    Hii @Aladdin Persson. Amazing video. I just have a doubt. While calculating iou for true_label and pred_labels, should we not add the width and height that we clipped when creating true_labels? That is, in case of the example you gave of [0.95, 0.55, 0.5, 1.5], shouldn't we convert 0.95 to 0.95(as the cell we chose is in 0th index along the width) and 0.55 to 1.55(as the cell we chose is in 1st index along the height). This is because we are doing geometric operations like converting x_centre and y_centre to xmin, ymin, xmax and ymax and on not doing the conversion I mentioned, instead of getting the xmin, ymin, xmax and ymax of the bounding box we get some other coordinates instead.
    Also could you please create the same using Tensorflow?

  • @zukofire6424
    @zukofire6424 Год назад

    Thanks! I don't understand the code regarding the bounding boxes though... Could you do a deep dive into the bounding boxes calculations AND show how to test on a new image?

  • @mahdiamrollahi8456
    @mahdiamrollahi8456 2 года назад +1

    Hello. Why the target and prediction are in different shapes?

  • @horvathbalazs1480
    @horvathbalazs1480 4 года назад +3

    Hi, I really appreciate your work and patience to make this video, however I would like to ask the following: The loss function is created based on the original paper, but the loss for bounding box midpoint coordinates (x,y) are not included because we calculate just the sqrt of width, height of boxes. Am I right?

    • @horvathbalazs1480
      @horvathbalazs1480 4 года назад +3

      Okay, sorry for the silly question. I just noticed that we should not get the squared root of x,y so that's why we skip here:
      box_predictions[..., 2:4] = torch.sign(box_predictions[..., 2:4]) * torch.sqrt(
      torch.abs(box_predictions[..., 2:4] + 1e-6)
      )
      box_targets[..., 2:4] = torch.sqrt(box_targets[..., 2:4])

  • @yantinghuang7491
    @yantinghuang7491 4 года назад +1

    Great video! Will you make "from scratch" series video for Siamese network?

    • @AladdinPersson
      @AladdinPersson  4 года назад

      I'll look into it! Any specific paper?

    • @yantinghuang7491
      @yantinghuang7491 4 года назад

      @@AladdinPersson Thanks Aladdin! This one should be a good reference: Hermans, Alexander, Lucas Beyer, and Bastian Leibe. "In defense of the triplet loss for person re-identification." arXiv preprint arXiv:1703.07737 (2017).

  • @R0Ck50LiD-b5z
    @R0Ck50LiD-b5z 2 года назад +1

    Hi, do you have any details on how you prepared the dataset?

  • @siddhantjain2591
    @siddhantjain2591 4 года назад +2

    Awesome as always!
    Could you do some video on EfficientNets sometime, that would be great !

  • @markgazol5404
    @markgazol5404 4 года назад +2

    Very clear and helpful! Thanks for the videos. I've got one question, though, Can you please explain what is the label for the images with no objects? During the training should it be like [0, 0, 0, 0, 0] or smth?

  • @saeeddamadi3823
    @saeeddamadi3823 3 года назад

    At 1:05:41 you mention your video of how to build a custom dataset. Please link it to the video to enhance your informative channel.

  • @kayleescanlin4699
    @kayleescanlin4699 5 месяцев назад

    Can someone explain what the "conda activate dl" means in the terminal at 57:27? Is that a specific environment to download or is it something we create ourselves?

    • @SAnish-uj4jc
      @SAnish-uj4jc 5 месяцев назад

      yo im not able to understand the code ?? am i missing something please help

  • @Sky-nt1hy
    @Sky-nt1hy 3 года назад

    There's an error at 56:11, line 10 : it Should be target[..., 25:26] instead of target[..., 20:21] for the no-object-detected loss

    • @정현호-u9j
      @정현호-u9j 3 года назад

      I think target[...,20:21] is right. The target's index for last dimension ends at 24, I think

    • @Sky-nt1hy
      @Sky-nt1hy 3 года назад

      @@정현호-u9j 아하 그렇군요! 감사합니다ㅎㅎ

    • @정현호-u9j
      @정현호-u9j 3 года назад

      제가 이해하기로 target data 값은 7x7x25 의 shape을 갖어요. 25중 앞의 20은 각 01000.. 처럼 해당 class가 one hot encoded 된 값이고. 나머지 5는[ confidence score, x,y,height,width]인데, 왜 5*2가 아니라 5냐면 논문에서는 하나의 grid cell에는 하나의 true object( 정확히는 object를 감싸는 bounding box의 mid point)만 존재한다고 가정하고 있어요. 혹시 제가 틀렸다면 고쳐주세요

  • @RicardoRodriguez-nn5jw
    @RicardoRodriguez-nn5jw 4 года назад

    Hey man i just found your channel, really good videos. I just saw that you are doing also a tensorflow playlist, are you planning to make maybe a yolo3,4 on tensorflow like this one from pytorch? Maybe common implementations, yolo or mtcnn, pcn?
    Looking forward to it! Greeeeets

  • @wuke4231
    @wuke4231 Год назад

    thank you for your video!😘

  • @hichensstark1048
    @hichensstark1048 4 года назад

    i have wathed all if the videos !!!

  • @larafischer420
    @larafischer420 Год назад +1

    muito boa essa série de vídeos! Vc pode passar as referências q vc usa pra montar esses notes? Tenho dificuldade em encontrar materiais pra estudar

  • @danlan4132
    @danlan4132 2 года назад

    Thank you very much!!!! Excellent video!!!! By the way, do you have any tutorials for oriented bounding box detection?

  • @nerdyguy7270
    @nerdyguy7270 2 года назад +2

    Hi, this is awesome and really helpful. I was going through the yolov1 paper and found that the height and the width are relative to the whole image and not to the cell. Is that correct?

  • @josephherrera639
    @josephherrera639 4 года назад +3

    Do you mind showing how to plot the images with their bounding boxes (and how that can be applied to testing on new data)? Also, do all images have a maximum of 2 objects to localize?

  • @talhayousuf4599
    @talhayousuf4599 3 года назад

    Too much Thanks for this video, I'm anxiously waiting for Yolo v3 . Can you pleaseee.... do such video for that ?

  • @omarhesham7390
    @omarhesham7390 8 месяцев назад

    Fantastic Bro

  • @anshulgoyal1095
    @anshulgoyal1095 3 года назад

    Works well on Colab GPU. Just need to change the addresses of file references.

  • @Wh1teD
    @Wh1teD 3 года назад +1

    Very informative video and I think I understood the algo but there is one doubt I have: the code you wrote would only work with this specific dataset? If I would want to use a different dataset, would I need to rewrite the bigger part of the code (i. e. the loss function, the training code)?

  • @apunbhagwan4473
    @apunbhagwan4473 3 года назад +1

    He is simply Great

  • @amartyabhattacharya
    @amartyabhattacharya Год назад

    One question that I have is, How can I get to know the coordinates of the grid cell of which the centers are a part of? Is it like (1,1) of the output prediction gives the prediction for grid cell having two endpoints as (0,0),(64,64) ? (448/7 = 64)

  • @dengzhonghan5125
    @dengzhonghan5125 2 года назад

    Thanks for your awsome video which really helps me understand the concept. (code always tell us the truth)