ฝัง
- เผยแพร่เมื่อ 4 ก.ค. 2024
- ✅ Subscribe: bit.ly/rf-yt-sub
A video of how to train YOLO v4 to recognize custom objects in Google Colab in the Darknet framework. In this video we will take the following steps to train our custom detector:
1) Gather and process our dataset
2) Load dataset into Google Colab
3) Build Darknet framework in Google Colab
4) Write custom YOLO v4 training configuration
5) Train custom YOLO v4 detector
6) Use trained YOLO v4 detector for inference
7) Export YOLO v4 weights
Label your images:
blog.roboflow.com/getting-sta...
Corresponding training blog post:
blog.roboflow.com/training-yo...
Colab Notebook:
colab.research.google.com/git...
What is MaP?
• What is Mean Average P... - วิทยาศาสตร์และเทคโนโลยี
I have to admit, when I first started watching this I thought to myself "ugh, another AWS overpriced solution". But no, this is actually awesome. You've saved me a ton of hours in getting the data ready for our own models - new subscriber AND customer!
Thanks for the great videos.
you Rock!!!
Thanks a lot for allowing so many to get their feet wet with visuals.
Just starting of with integrating it into our existing solutions.
QUESTION:
How do I change/reduce the file size of the generated .weigths file?
Which factores do control it and where can I change them?
Hi, thanks for the great explanation. Is there any chance that we can see the accuracy of the model (YOLO v4) after the training process done? thank you
Thanks you share the procedure,.
I've the question is that could I use the weights on video?
Hi! Thank you for this tutorial. Can you tell me how can I view/download my annotation labels in txt format? I used roboflow to generate my dataset and it created txt files for every image in the test directory but they remain empty even after testing my detector. I need the generated darknet format labels for my test set.
Once we've got our (trained) best weights, how can we use them on unseen data of the same kind just for testing? And can inference be tested on multiple images simultaneously?
Thank you very much!
very helpful sir for my project. upload more cool videos
Hi! Thank you for the explanation. I have a few questions. When you download the pre-trained weights, is this called a transfer learning? What's the difference between transfer learning and fine-tuning? Is there a way to show the mAP or loss in graph format? Sorry I'm still new to this..
bang kuliah di mana
Hi! Question, I have exported my weights if I want to predict using this notebook still how can I do that, I already imported the weights but it does not predict sadly.
@Robotflow Great video! How can i get the coordinates of the predicted boundboxes?
Thank you so much
I'm training the yolov4 to identify small objects, these objects are one into another... Im identifying cells and nuclei.... When i finished the process, the name of the bounding boz were to big and they were above the other ones... it became impossible to read. Is there any chance the make the letters smaller to see the others predictions?
thanks you are the best
Goof afternoon sir...sir I would like to ask on how you installed your pip requirements using google colab,,because I followed your tutorial but I have encounter some error in converting yolo to tflite model
thanks for the great video
how did u change the image @19:23 ?
Can I ask
what is the number of gpu tesla V-100 on colab? :)
Thank you for your easy to follow videos:). Please I have a question. Why do we have to pre-structure the train and validation images before running the training? Can't we implement k-fold cross validation on yolo during the training?
Most of the tutorials that i have found in yolo whether v3, v4 or v5 has the train and validation folders already setup. Thanks for your response. Am really curious about this.
Yes, you should have a validation set already split apart! In the video you can see how Roboflow will automatically split your data for you.
@@Roboflow Yes i saw in the video, that the split was done by Roboflow.
Is it possible to do the split using k-fold Cross validation during training rather than using Roboflow for the split?
@@iposipos9342 k-fold cross validation may indeed be possible - though it is not in the scope of this tutorial.
@Roboflow Should I training the model again if I make augmentation on the dataset, Where the model has been trained on an original dataset before?
Yes. Consider starting training from the checkpoint of the model trained on your original images.
can we detect the accident dent using yolov4 in darknet? if yes then how & if no then please provide option!
When performing inference should I resize the input to match the training size, or does darknet do that automatically? Does it resize training input images if they’re not already the correct size?
Yes the images will be automatically resized. However, you may see slightly faster inference if you resize ahead of time.
How to upload the data directly without using your Roboflow app. Because in Roboflow, I cannot split my train, Val, and test sets category-wise if the data is imbalanced. Not sure how your build custom model did not take that into consideration.
hello bro, can we use tensorboard for yolov4?
Hi I have some problems I use custom data on roboflow when I'm training it have errors path directory not have ,how i can fix this problem
can you do age prediction by facial feature recognition using yolo v4
how to save files with .labels and .cfg extensions in collab ?
now the colab has changed, the code inside the colab notebook provided unable to run completely already. Is there anyone is currently doing yolov4 with this colab notebook but can run completely ?
how to change the number of epochs and then create checkpoints to save those trained weights
Isn't it too slow for training? Because only 50 photos for train, validation and test. How can I do faster?
Hi Thanks for the video. I got this error.
CUDA Error Prev: an illegal memory access was encountered
CUDA Error Prev: an illegal memory access was encountered: Success
darknet: ./src/utils.c:325: error: Assertion `0' failed.
I am getting the below error when I have to perform the prediction
Unable to init server: Could not connect: Connection refused
It will be really helpful if you add code for testing .mp4 or video , hope you will add this in your collab repo
!./darknet detector demo data/obj.data cfg/custom-yolov4-detector.cfg backup/custom-yolov4-detector_last.weights -dont_show /content/traffic.mp4 -thresh 0.4 -i 0 -out_filename /content/traffic_1.mp4
add this line instead of the one provided for image inference and replace all the cfg and weights with yours in this command including the video path.
need the best file in py. How can I do that .
Its still work on 2023 ?
@Roboflow, I am unable to run the below part, getting an error stating cannot find _darknet.labels even though I followed your entire code. Could you please help me with this?
#Set up training file directories for custom dataset
%cd /content/darknet/
%cp train/_darknet.labels data/obj.names
%mkdir data/obj
#copy image and labels
%cp train/*.jpg data/obj/
%cp valid/*.jpg data/obj/
%cp train/*.txt data/obj/
%cp valid/*.txt data/obj/
with open('data/obj.data', 'w') as out:
out.write('classes = 2
')
out.write('train = data/train.txt
')
out.write('valid = data/valid.txt
')
out.write('names = data/obj.names
')
out.write('backup = /mydrive/yolov4/backup')
#write train file (just the image list)
import os
with open('data/train.txt', 'w') as out:
for img in [f for f in os.listdir('train') if f.endswith('jpg')]:
out.write('data/obj/' + img + '
')
#write the valid file (just the image list)
import os
with open('data/valid.txt', 'w') as out:
for img in [f for f in os.listdir('valid') if f.endswith('jpg')]:
out.write('data/obj/' + img + '
')
Make sure to export in the "YOLO Darknet" format!
@@Roboflow , yes I did...exported.
i have taked erorr: "/bin/bash: ./darknet: No such file or directory"
Please create an issue at github.com/roboflow-ai/notebooks and describe your problem. We will try to help you out :)
have you fixed the error ?
@@rusugabriel2554, unfortunately no, i stopped using roboflow, here is my solution :)
I have done all the steps but when I am trying to test the images I amgettiing "test" folder is not there. Has anyone faced similar issue?
FileNotFoundError Traceback (most recent call last)
in
1 import os
2 #/test has images that we can test our detector on
----> 3 test_images = [f for f in os.listdir('test') if f.endswith('.jpg')]
4 import random
5 img_path = "test/" + random.choice(test_images);
FileNotFoundError: [Errno 2] No such file or directory: 'test'
In my case I created a completly new dir for the test images but I think the issue was that the directory in the code did not match mine so I changed mine to /content/drive/MyDrive/ColabNotebooks/test
I've got 80 images, whys it say it's gonna take 10 hours? :O
It doesn't function anymore.
how to do it? checkpoint,yolov4.data-00000-of-00001,yolov4.index ----> saved_model.pb
/bin/bash: line 1: ./darknet: No such file or directory
i have the same problem, did you found the solution?
during 16:24 steps. You need to go outside and touch the grass. Because this step take so ducking long.
Yeah... the title was about training in darknet. Dislike
i got this error "error: ‘CUDNN_CONVOLUTION_FWD_PREFER_FASTEST’ undeclared (first use in this function); did you mean ‘CUDNN_CONVOLUTION_BWD_FILTER_ALGO_3’?" when i run "!make". Someone knows how to solve this?
i got issue on 10:14
compilation terminated due to -Wfatal-errors.
make: *** [Makefile:144: obj/convolutional_layer.o] Error 1
why i get this error when compile darknet using Cmake : compilation terminated due to -Wfatal-errors.
make: *** [Makefile:162: obj/convolutional_layer.o] Error 1
Hi 👋🏻! It is a known bug that we are trying to solve for quite a long time. You can take a look at the issue here: github.com/roboflow/notebooks/issues/28
Thank you so much