blcakhat / yolov3-helmet-detection Goto Github PK

Hey i'm having this issue, line 37:

return [layersNames[i[0] - 1] for i in net.getUnconnectedOutLayers()]

Hi,
Is this model 'yolov3-obj_2400.weights' only able to detect helmet? I wonder how many classes can be detected.
The reason I ask this question is if the model 'yolov3-obj_2400.weights' can detect helmet as well as person, we can further judge whether a person is wearing a helmet or not, just like liveness detection.
Thank you!

I am sorry but I don't know how this model can be used to find the people ridding bike without helmet. Could you please explain it to me? Because I just see the label "helmet" of your model, so if I want to detect without helmet, I need to retrain it, right?

I am geting the following error when I am running this : python Helmet_detection_YOLOV3.py

Traceback (most recent call last):
File "Helmet_detection_YOLOV3.py", line 28, in
net = cv.dnn.readNetFromDarknet(modelConfiguration, modelWeights)
AttributeError: 'module' object has no attribute 'dnn'

My python version is: Python 2.7.9
Cv2 version: 2.4.9.1

Hi, and thanks for uploading those files.
due to problem using darknet directly, I'm using a tf translation for the dark net.
it has a convertor - from weights file to .tf file, so the net could use.

Your weight file (and another one I got from kaggle), I got dim problem - can't reshape an array of size 4607 into shape (18,256,1,1) (which is 4608 parameters).

Do you have any idea what could be the source of the problem?
Other weight files I use are doing just fine.

Thank you for answering,
Tomer

i want to use video or streamin the input and detection helmets ,

How can we detect helmetless travellers on a bike using this repo

Hi @BlcaKHat ,
I'm not able to download the model file provided by you. It's showing the model file is in your trash.
Can you share the same on [email protected] please?
Thanks.

Hi,

Please have a look at my project structure : -

https://ibb.co/fnfKPBn
https://ibb.co/88rjZCB

Im getting None as output for all the images.
Do I need to install darknet on windows ? But it isnt giving any error regarding missing darknet or any other error for that matter.
OS : Windows10

Hi,
could you have train your custom object successful on google colab ?
Best Regards

Thanks for your contribution! But where to download the dataset?

Hello @BlcaKHat, what is the source for your helmet dataset? I currently use your custom weight and it seems very good. I wanna try it with yolov4 but with my own dataset from openimage dataset is not quite good as this one. So I wanna know how many pictures you use and what is the source of your dataset.

Thank you

I wonder to know where can I get the helmet training dataset that can be used to YOLOv3
Thanks a lot

hi, my name is manuel.
I'm trying to run the example.
i have downloaded the git repo and the yolo weigs, cfg and names.
Changed the paths to files as your instructions in the readme

when tryng to run, in the readnetfromdarknet i got
Error: Parsing error (Unknown layer type: shortcut) in cv::dnn::darknet::ReadDarknetFromCfgFile, file C:\Miniconda3\conda-bld\opencv-suite_1533128839831\work\modules\dnn\src\darknet\darknet_io.cpp, line 503
PS C:\Source\Pruebas Deteccion\yolov3-Helmet-Detection-master>

i have veryfied the file paths.

Opencv 3.4.1
Python is 3.7.4
Windows 10

thanks

Hi,
Which IDE are you using?
Best regards,
PeterPham