Thanx for making this channel, because this kind of project really helps beginners, your coding is pure gold and your teaching is pure Diamond. Once again thanx for the help, hope you will make another live project in nearly future.
hey im encountering various compatibility issues with the new tensorflow version and object detection api so can you make a video with the latest tensorflow version
Great work Nick ! I had followed your steps but I am getting a higher training loss even after training 50000 steps. Sample Log: I1203 13:38:15.593827 1752 model_lib_v2.py:701] {'Loss/classification_loss': 1.6337632, 'Loss/localization_loss': 0.6594347, 'Loss/regularization_loss': 5227525.5, 'Loss/total_loss': 5227528.0, 'learning_rate': 7.710457e-05} INFO:tensorflow:Step 49800 per-step time 0.126s I1203 13:38:28.178768 1752 model_lib_v2.py:698] Step 49800 per-step time 0.126s INFO:tensorflow:{'Loss/classification_loss': 1.1009775, 'Loss/localization_loss': 0.70923495, 'Loss/regularization_loss': 5227524.0, 'Loss/total_loss': 5227526.0, 'learning_rate': 3.426075e-05} I1203 13:38:28.184405 1752 model_lib_v2.py:701] {'Loss/classification_loss': 1.1009775, 'Loss/localization_loss': 0.70923495, 'Loss/regularization_loss': 5227524.0, 'Loss/total_loss': 5227526.0, 'learning_rate': 3.426075e-05} INFO:tensorflow:Step 49900 per-step time 0.126s I1203 13:38:40.809933 1752 model_lib_v2.py:698] Step 49900 per-step time 0.126s INFO:tensorflow:{'Loss/classification_loss': 1.1544539, 'Loss/localization_loss': 0.51572454, 'Loss/regularization_loss': 5227524.0, 'Loss/total_loss': 5227525.5, 'learning_rate': 8.559227e-06} I1203 13:38:40.816797 1752 model_lib_v2.py:701] {'Loss/classification_loss': 1.1544539, 'Loss/localization_loss': 0.51572454, 'Loss/regularization_loss': 5227524.0, 'Loss/total_loss': 5227525.5, 'learning_rate': 8.559227e-06} INFO:tensorflow:Step 50000 per-step time 0.126s I1203 13:38:53.439736 1752 model_lib_v2.py:698] Step 50000 per-step time 0.126s INFO:tensorflow:{'Loss/classification_loss': 1.3649243, 'Loss/localization_loss': 0.63460255, 'Loss/regularization_loss': 5227524.0, 'Loss/total_loss': 5227526.0, 'learning_rate': 0.0} I1203 13:38:53.442737 1752 model_lib_v2.py:701] {'Loss/classification_loss': 1.3649243, 'Loss/localization_loss': 0.63460255, 'Loss/regularization_loss': 5227524.0, 'Loss/total_loss': 5227526.0, 'learning_rate': 0.0} As a result not detecting the number plate from the image, not sure what went wrong. It will be helpful if u can advice.
thanks @Nicholas Renotte could you please help me with this error : but when trying run VERIFICATION_SCRIPT it replays with this error "module 'tensorflow' has no attribute 'contrib' "
Great video Nick! I love how you care about our queries and respond to each and every one! I have a small query too. How do I deploy the ml model we've made in jupyter notebook as a html website? Or a web app? Is it possible? Thank you.
A great video from you! it helps me a lot for my final year project! however, I need to add an extra feature that can detect the saved plate number that has other details in the database (or excel) so that the system able to display the detected plate number as registered (something like that). Do you know anything similar to this? I tried to find any similar to this but still haven't found yet. Thank you!
Hi Nicholas, very good video. I am trying to improve the performance of the program. I want to do is that when it reads a correct plate number, it skips the recognition for 2 seconds and after that time, the detection continues. Can you give mi any idea?.
Hello!Thank you for your teaching! Can I ask that ,when I use jupyter notebook to install tensorflow2.4.1,I get the error that could not find the version. I check that my python version is 3.7. Why I still get wrong?
Hi! Great Tutorial! But how can I use this model as part of my project? What if other want to run my project? Will they have to download and perform all these steps? Is there an easier way to quickly use the model?
Hi Nick, thanks for the tutorial. It's wonderful❤ anyway, I have a question. Why does it take longer time when detect & recognize in real time? Any spesific reason?? I'd be happy if you give me an explaination for that. And for those who read my comment & know the answer, please reply this🤗😊. Thanks & have a nice day!
Hey Nick.. I love watching your videos! So while running this I ran into a strange error while importing easyocr. OSError: [WinError 127] The specified procedure could not be found. Error loading "D:\ANPRS\anpr\lib\site-packages\torch\lib\cudnn_adv_infer64_8.dll" or one of its dependencies. All the steps ran smoothly up till 1:13:50 I do have the exact dll file in the location, also tried restarting the kernel and even tried installing other torch versions. The same error pops up. Any suggestions?
Hello Nicolas,Great video and content. Please can you help me, i am getting an error while running !(command) cell ,am trying to solve from past 10 days: Traceback (most recent call last): File "Tensorflow\models esearch\object_detection\model_main_tf2.py", line 32, in from object_detection import model_lib_v2 ModuleNotFoundError: No module named 'object_detection' Restarted the kernel, then also was showing the error, have also installed cuda -10.1 and cudnn -7.6.5 for tfod api.
I am getting this error - File "/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/load_library.py", line 178, in register_filesystem_plugin py_tf.TF_RegisterFilesystemPlugin(plugin_location) tensorflow.python.framework.errors_impl.AlreadyExistsError: File system for s3 already registered
Hi During 10. Real Time Detections from your Webcam, many error codes appear. ValueError: 'images' must have either 3 or 4 dimensions. how do i fix this ?? uses USB web camera.
Could not find a version that satisfies the requirement tensorflow==2.4.1 (from versions: 2.12.0rc0, 2.12.0rc1, 2.12.0, 2.13.0rc0) ERROR: No matching distribution found for tensorflow==2.4.1 i am getting this error
Hi, How to get specific colored number plates recognisation, means if I want a model to recognize only yellow or white number plates so I will achieve this goal
@@NicholasRenotte if I took a shot for every error i got i would now be i drunk haha. I get this weird error `AttributeError: module 'tensorflow' has no attribute 'contrib'` after running the test, i installed tensorflow 2.4.1 but it just keeps giving errors in the VERIFICATION_SCRIPT. i installed tensorflow both in the virtual environment and the environment outside but when i try to do it via cmd it says `ERROR: No matching distribution found for tensorflow==2.4.1` I've been stuck on this for 2 hours now, do you have any idea how to fix this?
Firstly , wow, i found your channel yesterday and its PERFECT. Love the detail, no stone left unturned, really clear, thank you for taking the amazing amount of time i must take to put these videos together. I instantly worked through this video step by step when i found it. Initially i struggled to get it working, i tried it on Windows, Ubuntu and got it working on a Mac. Each OS gave different errors when trying to get the right versions. I have a Coral TPU, I'll try and get it working with that . Thanks again, seriously awesome tutorial.
great video nick, i have gotten to # 9. Detect from an Image but my image is displaying dull and it not showing the percentage text, but i my computer doesn't support NVIDIA.
Hi Nick, thanks for the tutorial. When I use jupyter notebook to install tensorflow 2.4.1, I get the error that could not find the version. My version of python is 3.7. When I use the tensorflow==2.10.0 version, there are 2 errors: 1. Downgrade the protobuf package to 3.20.x or lower. 2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower). When I install protobuf package to 3.20 there is an error: ImportError: cannot import name 'builder' from 'google.protobuf.internal' etc. How I solve this problem?
Hey Nick, thanks so much for the great video! Just wondering how I would go about exporting the trained Tensorflow model to be used in another Python file? Basically I just need to be able to detect car plates so I wanted to know if I would be able to export the Tensorflow model and use it as a function maybe?
Hi Nick. I'm very new to this but interested in exploring. I presume I need to setup the Nvidia CUDA drivers on my laptop ( I have a Microsoft Surface Book 3 with an Nvidia GeForce GTX 1660 Ti ) in order run trainings. Do you have any videos on setting up the CUDA Drivers on a Windows 10 laptop or can pass on some links you'd recommend? And thank you for sharing and making this channel. I enjoy your style and attention to detail. Cheers and Merry Holidays!
Awesome work @nicholasrenotte. Took a while to fix all the errors i encountered mainly because I am a core javascript dev. Once it was working as expected I exported my saved model into tfjs and integrated with my react app (All thanks to your 5 hour long video)
@@TahaNourAkhElArab Unfortunately for me , I encountered pip dependency issues particularly with tensorflow packages that took me a couple of days to fix.
That was a great session! I want your advice on how to deploy easyOCR on mobile devices. I couldn't find any resources on this and it would be great to point me at the right direction. Thanks!
Hi Nick, I'm looking to run this in a Raspberry Pi edge AI camera. Is there an export function for tensorflow in the same way as yolo provides or would you recommend yolo in this case?
Is it possible for implementation of the software on the front end, meaning is there anyway in which we can add a front end interface to it, like a website or something and can we include the same code and data and hide them in the backend. If possible please let me know and also please let me know how I can implement the same?
Hmmm, just took a quick look. There aren't annotations for the Stanford dataset, you'd probably be better off using classification for that particular dataset @Ruchita!
I started training the model with 50 steps, it has been training for more than 7 hours and it is still not in that training process. Is it normal that it has been training for so long with 50 steps or is it due to an internal error that is not shown in the logs?
Hey appreciated the help for my last problem. But for now, as soon as I want to start the training, there's an error saying 'property' object is not callable. Is it because I'm using a different device? I've matched the tensorflow version and most of other libraries. Thanks in advance
Thanx for making this channel, because this kind of project really helps beginners, your coding is pure gold and your teaching is pure Diamond. Once again thanx for the help, hope you will make another live project in nearly future.
Thanks so much @Shivam, so glad you liked it!
The people who are getting errors let me tell you ,it work ,it work for me you just have to spend some time on them ...thankyou nicholas
hii
hey im encountering various compatibility issues with the new tensorflow version and object detection api so can you make a video with the latest tensorflow version
Great work Nick ! I had followed your steps but I am getting a higher training loss even after training 50000 steps.
Sample Log:
I1203 13:38:15.593827 1752 model_lib_v2.py:701] {'Loss/classification_loss': 1.6337632,
'Loss/localization_loss': 0.6594347,
'Loss/regularization_loss': 5227525.5,
'Loss/total_loss': 5227528.0,
'learning_rate': 7.710457e-05}
INFO:tensorflow:Step 49800 per-step time 0.126s
I1203 13:38:28.178768 1752 model_lib_v2.py:698] Step 49800 per-step time 0.126s
INFO:tensorflow:{'Loss/classification_loss': 1.1009775,
'Loss/localization_loss': 0.70923495,
'Loss/regularization_loss': 5227524.0,
'Loss/total_loss': 5227526.0,
'learning_rate': 3.426075e-05}
I1203 13:38:28.184405 1752 model_lib_v2.py:701] {'Loss/classification_loss': 1.1009775,
'Loss/localization_loss': 0.70923495,
'Loss/regularization_loss': 5227524.0,
'Loss/total_loss': 5227526.0,
'learning_rate': 3.426075e-05}
INFO:tensorflow:Step 49900 per-step time 0.126s
I1203 13:38:40.809933 1752 model_lib_v2.py:698] Step 49900 per-step time 0.126s
INFO:tensorflow:{'Loss/classification_loss': 1.1544539,
'Loss/localization_loss': 0.51572454,
'Loss/regularization_loss': 5227524.0,
'Loss/total_loss': 5227525.5,
'learning_rate': 8.559227e-06}
I1203 13:38:40.816797 1752 model_lib_v2.py:701] {'Loss/classification_loss': 1.1544539,
'Loss/localization_loss': 0.51572454,
'Loss/regularization_loss': 5227524.0,
'Loss/total_loss': 5227525.5,
'learning_rate': 8.559227e-06}
INFO:tensorflow:Step 50000 per-step time 0.126s
I1203 13:38:53.439736 1752 model_lib_v2.py:698] Step 50000 per-step time 0.126s
INFO:tensorflow:{'Loss/classification_loss': 1.3649243,
'Loss/localization_loss': 0.63460255,
'Loss/regularization_loss': 5227524.0,
'Loss/total_loss': 5227526.0,
'learning_rate': 0.0}
I1203 13:38:53.442737 1752 model_lib_v2.py:701] {'Loss/classification_loss': 1.3649243,
'Loss/localization_loss': 0.63460255,
'Loss/regularization_loss': 5227524.0,
'Loss/total_loss': 5227526.0,
'learning_rate': 0.0}
As a result not detecting the number plate from the image, not sure what went wrong. It will be helpful if u can advice.
overfitting vro!
Much awaited video... will check out the whole thing soon ;))
Awesome, let me know how you go @Godfrey!
@@NicholasRenotte hey Nick did you use yolo to detect license plate or created your own model
8:25, folder called ANPR nothing too crazy there right, its just a folder
haha made my day
This is surely going on my resume
Amazing Work Dude!!!
Is this a Machine Learning project??
Sure is!
finally i completed this project!🥳
thanks @Nicholas Renotte
could you please help me with this error :
but when trying run VERIFICATION_SCRIPT it replays with this error "module 'tensorflow' has no attribute 'contrib' "
It was my first "project" and i want to say THANK YOU Nicholas ! It was not easy but I learned a ton on the way !
Is this code work?
Hey Mann!! Found ur channel pretty recently. Ur content is 🔥🔥🔥 Thanks for delivering such amazing content!!
AYYYY, welcome to the channel!
Initially when we are supposed to follow all the steps within the github repository are we supposed to have Python installed on the system?
Great video Nick! I love how you care about our queries and respond to each and every one! I have a small query too. How do I deploy the ml model we've made in jupyter notebook as a html website? Or a web app? Is it possible? Thank you.
I try to get to them all @Tejeshwar. Sure can, check this out, it shows you how to do it exactly: ruclips.net/p/PLgNJO2hghbmjNrHZqplNMEpsW-QLFdvJV
With you, my life is becoming easier.. Brother... All the best
Awesome! Thanks so much @Priyam, planning on building it?
@@NicholasRenotte yep yep. So right now I am building the OpenCv + cuda acceleration.
@@priyambordoloi771 awesome!
Thank you!!
Amazing work!!
Thank you soo much @Vendroid!
which is this python version?
Please reply
Can we use this as our final year project?
Definitely!
Thank you very much for this video. It is very detailed and easy to follow.
A great video from you! it helps me a lot for my final year project! however, I need to add an extra feature that can detect the saved plate number that has other details in the database (or excel) so that the system able to display the detected plate number as registered (something like that). Do you know anything similar to this? I tried to find any similar to this but still haven't found yet. Thank you!
Not exactly but you could definitely integrate into other systems from the detected plate @Nuha, possibly using a SQL query or something :)
@@NicholasRenotte alright, thank you ! :)
hey i also need to give this project in my end semester nd time is less will you please help me in report nd ol
@Nuha Syuhada can I have email. I want to ask some questions regarding it
Hey can i alsoget your email i have questions regarding the database
Thank you Bhai for helping me with my project! Bappa Bless You!
haha took my computer almost 5 hours to train
You are now a disciple of ML. Lol I’m training a game rl model and it’s been 51 hours now.
@@NicholasRenotte can i run train command multiple time ? and what will happen ?
@@arazkrdsorry your computer is no more 😂
Lolz. Will NVidia 4060 take less than 5 hr ? Lolz
Hi Nick, is there an implementation of this project in Google collab ?
im getting error how to resolve it?
module 'tensorflow' has no attribute 'contrib'
Try reinstalling, this normally means that the contributor components of TF are missing.
did you solve it?
@@davidgrimm4063 no but I tried to make the code and it's working
Hi Nicholas, very good video. I am trying to improve the performance of the program. I want to do is that when it reads a correct plate number, it skips the recognition for 2 seconds and after that time, the detection continues. Can you give mi any idea?.
Hello!Thank you for your teaching! Can I ask that ,when I use jupyter notebook to install tensorflow2.4.1,I get the error that could not find the version. I check that my python version is 3.7. Why I still get wrong?
did u find the solution
Did you find the solution?
Hi! Great Tutorial! But how can I use this model as part of my project? What if other want to run my project? Will they have to download and perform all these steps? Is there an easier way to quickly use the model?
Hi Nick, thanks for the tutorial. It's wonderful❤ anyway, I have a question. Why does it take longer time when detect & recognize in real time? Any spesific reason?? I'd be happy if you give me an explaination for that. And for those who read my comment & know the answer, please reply this🤗😊. Thanks & have a nice day!
Might be due to you're machine :), running this on real time is super computationally expensive.
@@NicholasRenotte thanks for the answer Nick!😄
How to solve 'images' must have either 3 or 4 dimensions error?
Hey Nick.. I love watching your videos!
So while running this I ran into a strange error while importing easyocr.
OSError: [WinError 127] The specified procedure could not be found. Error loading "D:\ANPRS\anpr\lib\site-packages\torch\lib\cudnn_adv_infer64_8.dll" or one of its dependencies.
All the steps ran smoothly up till 1:13:50
I do have the exact dll file in the location, also tried restarting the kernel and even tried installing other torch versions. The same error pops up. Any suggestions?
Heya! Double check you're using the appropriate version of PyTorch for your cudnn installation (in this case, it looks like cudnn 8)
Hi Nick, I am getting an error when running verification script:
AttributeError: module 'tensorflow' has no attribute 'contrib'
Hi Nick I am receiving the same issue. Please look into this when you get the chance!
sameeeeee
Hi Nicholas. Can I get one for pycharm
I am getting the error "No module named TensorFlow."My system OS is Windows 10, with AMD Radeon Vega 8 graphics. How should I fix it?? Pls tell soon
U have to install tensorflow
dud if u dont know how to fix that i think u shouldnt try to make this cuz u will get hell lot of err messages worse than that
Many thanks to this GOOD video! It helps me a lot
🙏🙏🙏
Hello Nicolas,Great video and content. Please can you help me, i am getting an error while running !(command) cell ,am trying to solve from past 10 days:
Traceback (most recent call last):
File "Tensorflow\models
esearch\object_detection\model_main_tf2.py", line 32, in
from object_detection import model_lib_v2
ModuleNotFoundError: No module named 'object_detection'
Restarted the kernel, then also was showing the error, have also installed cuda -10.1 and cudnn -7.6.5 for tfod api.
Run pip list from your notebook and ensure that object_detection is displayed there!
@@NicholasRenotte Thanks, I closed the Jupiter Notebook, again started it, then it worked
Hello Sir, I am getting error at step 5 while doing update config. Type error : Descriptors cannot be created directly it's coming. Please help me out
I am getting this error -
File "/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/load_library.py", line 178, in register_filesystem_plugin
py_tf.TF_RegisterFilesystemPlugin(plugin_location)
tensorflow.python.framework.errors_impl.AlreadyExistsError: File system for s3 already registered
i'm too
me too ,did u guy's find the solution
@@vaibhavthapli7216 Nope
Thank you so much for making this. You're a great tutor.
Hey do you get tar and protoc is not recognised as internal and external command
@@pinkchannel14 No, it worked perfectly for me
Hi
During 10. Real Time Detections from your Webcam, many error codes appear. ValueError: 'images' must have either 3 or 4 dimensions. how do i fix this ?? uses USB web camera.
Likely the webcam device number is incorrect, try an alternate id.
Hey NIcholas, How effective is EasyOCR is low light conditions. Do you have anyone other OCR or does this perform a good job?
Yep, PaddleOCR works pretty well as well: github.com/PaddlePaddle/PaddleOCR
Could not find a version that satisfies the requirement tensorflow==2.4.1 (from versions: 2.12.0rc0, 2.12.0rc1, 2.12.0, 2.13.0rc0)
ERROR: No matching distribution found for tensorflow==2.4.1
i am getting this error
Hi, How to get specific colored number plates recognisation, means if I want a model to recognize only yellow or white number plates so I will achieve this goal
Absolutely legend 🙏
Thanks so much @Nandan!
How can i get only the correct answer to be stored , because i need to use this in a project that saves the plates and the date . Help me please😢
Bro which version of python i would i choose for this project bro i am confused bro please
it's ok if use multiple dataset right? what I mean can use a dataset for cars and another dataset for persons in the test and train folder?
Hmmm, yes but you would need an additional condition in the detection to check if the detected class is a plate when running the models!
Thanks brother great project.❤
Dude you are just awesome 😀😀
Ohh shucks, thanks so much @Rahul!!
Awesome video dude!! Really clear!
I need to make exactly this for school so this will help me out a lot!
Perfect! Let me know how you go with it @Jaap!
@@NicholasRenotte if I took a shot for every error i got i would now be i drunk haha.
I get this weird error `AttributeError: module 'tensorflow' has no attribute 'contrib'` after running the test, i installed tensorflow 2.4.1 but it just keeps giving errors in the VERIFICATION_SCRIPT.
i installed tensorflow both in the virtual environment and the environment outside but when i try to do it via cmd it says `ERROR: No matching distribution found for tensorflow==2.4.1`
I've been stuck on this for 2 hours now, do you have any idea how to fix this?
@@jaap6384 hmmm, what version of Python are you running?
@@jaap6384 heyyy!! hv you fix this error???
let me know pls
@@himanibansal445 not... after like 5 hours i ragequit and copied an already working repo from git and used that. sorry
Not proud of it either :)
hi, great video... nice way to explain things
Hey, Nicholas. I love your tutorials. Keep up the great work. Can you do a Jupyter Notebook tutorial?
Firstly , wow, i found your channel yesterday and its PERFECT. Love the detail, no stone left unturned, really clear, thank you for taking the amazing amount of time i must take to put these videos together. I instantly worked through this video step by step when i found it. Initially i struggled to get it working, i tried it on Windows, Ubuntu and got it working on a Mac. Each OS gave different errors when trying to get the right versions. I have a Coral TPU, I'll try and get it working with that . Thanks again, seriously awesome tutorial.
please help me. how do you install tf on macos? the guide only provide scripts for linux and macos. help!
@@myerwerl1 Hi, sorry I have only just seen this , did you get it working on the Mac ok?
Hey I'm trying to run it on Mac as well but facing subprocess issues while trying to install protobuf. Could u please help
What the python version did you use for this?
How should i go about getting more OCR results? should i process the image to make the quality increase or find a better OCR model/library?
Thankyou so much, can u please tell which algorithm have u applied to this project and where
We're using an OCR model and a Single Shot Detector for the Object Detection!
Is it comes under machine learning project?
Which OS version are you using???
For this, Windows!
Nick ,can I use it for my last year major Project.
Sure can!
What Neural Network are u using?
What version of python should we use?
Is that is possible that we use objtect detection to detecte the number instead of use easyocr?
Can you give me an advice Im planning to implement this on mobile devices .(is it possible)?
Wouldn't recommend it unless you have a ridiculously high powered mobile device!
great video nick, i have gotten to # 9. Detect from an Image but my image is displaying dull and it not showing the percentage text, but i my computer doesn't support NVIDIA.
hey @abudon5278. I am stuck on step. Can you help me out ?
Hi Nick, thanks for the tutorial. When I use jupyter notebook to install tensorflow 2.4.1, I get the error that could not find the version. My version of python is 3.7. When I use the tensorflow==2.10.0 version, there are 2 errors: 1. Downgrade the protobuf package to 3.20.x or lower.
2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower). When I install protobuf package to 3.20 there is an error: ImportError: cannot import name 'builder' from 'google.protobuf.internal' etc. How I solve this problem?
Hi, i am getting the same error. if u resolved it Can you please help
did you figure it out?
Hi, great video! Can you compare it with PaddleOCR?
Hey Nick, thanks so much for the great video! Just wondering how I would go about exporting the trained Tensorflow model to be used in another Python file? Basically I just need to be able to detect car plates so I wanted to know if I would be able to export the Tensorflow model and use it as a function maybe?
mutable default for field sgd is not allowed: use default_factory
Hi Nick, what is the name of the algorithm used in this program?
Hello There!
Quick Question what Python version are you using ?
Wow what a great video. Your video are always on point thanks for giving us such great work
Thanks so much @Mxolisi! So glad you enjoyed it!
I really did. Please don’t forget my 3D shoes detection using tflite for mediapipe
@@mxolisinkosinathinene2657 definitely, still got it planned!!
Okay thanks man I will be over the moon when it comes out
hi nicholas, great video as usual! anyway can i use this method to read power meter? (with different dataset as well)
Sure could!
@@NicholasRenotte great i don’t have to change the code right?
@@reylian.anareka nope, just train the object detector for your power meter!
Hi Nick. I'm very new to this but interested in exploring. I presume I need to setup the Nvidia CUDA drivers on my laptop ( I have a Microsoft Surface Book 3 with an Nvidia GeForce GTX 1660 Ti ) in order run trainings. Do you have any videos on setting up the CUDA Drivers on a Windows 10 laptop or can pass on some links you'd recommend? And thank you for sharing and making this channel. I enjoy your style and attention to detail. Cheers and Merry Holidays!
Awesome work @nicholasrenotte. Took a while to fix all the errors i encountered mainly because I am a core javascript dev.
Once it was working as expected I exported my saved model into tfjs and integrated with my react app (All thanks to your 5 hour long video)
Did you also have problems with version dependencies when you tried to shift enter folders on jupyter?
@@TahaNourAkhElArab I do
@@TahaNourAkhElArab So what do we can fix this bug ?
U had tried ?
@@TahaNourAkhElArab Unfortunately for me , I encountered pip dependency issues particularly with tensorflow packages that took me a couple of days to fix.
@@Ifeanyi4Ndu how do you solve them? I also encountered tensorflow dependencies issues.
That was a great session!
I want your advice on how to deploy easyOCR on mobile devices. I couldn't find any resources on this and it would be great to point me at the right direction.
Thanks!
Hmmm, probably reasonably involved. I know there's OCR capabilities on most phones so probably easier ways than what I've shown.
hey man thank you is this gonna work for boat plat recognition ??
Yep, just fine tune the detection model on boat plates @Idir!
The 2. Training and Detection ipynb notebook is loading still...i couldn't see any codes in my PC 19:10
is this from scratch as i am first time doing this so while following this will it work.?
Hi Nick! Thank you very much for a GREAT tutorial Can you tell which algo he is using ???
Object Detection and OCR!
Hi Nick, I'm looking to run this in a Raspberry Pi edge AI camera. Is there an export function for tensorflow in the same way as yolo provides or would you recommend yolo in this case?
this guy is too good!!!!!!!!!!
hi i m getting this error when i m running tranning command in cmd
cannot import name 'model_lib_v2' from 'object_detection'
can someone help me
Please check your working directory.
Hey Nick what version of python?
Is it possible for implementation of the software on the front end, meaning is there anyway in which we can add a front end interface to it, like a website or something and can we include the same code and data and hide them in the backend. If possible please let me know and also please let me know how I can implement the same?
Whats the difference between your last one project of ANPR
This one using a custom Object Detection model first to detect the plate which means it'll work a lot better with plates from different regions!
Wowweyy❤️ OCR!! I was looking for something similar for real-time barcode detection! Thanks as always!!
Anytime @Anisha, glad you enjoyed it!
That was really cool !!💯 Could you also show how we could write such codes in PyCharm using classes and functions ?
Thanks so much. Ya, might do it in a live stream thoughts?
@@NicholasRenotte Perfect !
@@rahiegadekar3422 you got it!
Is it suitable for beginners?
hello! I have a question, can I do this course with google colaboratory?
Hi Nick!
May I know how to run the real time detection in Google Colab at 1:42:15 ?
Can I use this technique for car make and model classification using kaggle stanford cars dataset. Can you help me ?
Hmmm, just took a quick look. There aren't annotations for the Stanford dataset, you'd probably be better off using classification for that particular dataset @Ruchita!
Great video - how do we package this into a usable application
Getting an error at 32:36:- name ‘PRETRAINED_MODEL_URL’ is not defined
Variable is defined towards the top @pravin :)
Please Make One more video on Face Based Recognition System. That's good for learning and keep it up your work is the world's best.
You got it!
nice tutorial. can you tell what is the python version, tensorflow version you used.
3.7.3 here :)
video was really good and helpful☺
Thanks a million @Varun! So glad you enjoyed it!
Indeed a great video Nick!!!! I am trying to implement the same in Google Colabs, can you please suggest me on how to do so?
Hi nick! Great video! I have a question, how could I change the ocr used in the code?
Should be able to plug in whichever OCR engine you need!
How can you apply it on facemask detection video that you created?
I started training the model with 50 steps, it has been training for more than 7 hours and it is still not in that training process. Is it normal that it has been training for so long with 50 steps or is it due to an internal error that is not shown in the logs?
Hey appreciated the help for my last problem. But for now, as soon as I want to start the training, there's an error saying 'property' object is not callable. Is it because I'm using a different device? I've matched the tensorflow version and most of other libraries. Thanks in advance
hey what is the maximum time that it takes after we type the training model command on the anaconda prompt
It shouldn't take more than 15 minutes to kick off even if you're starting off on CPU.
awesome video, thank you so much, so, i have a question is: the model can recog the vertical text?
Is this Category comes under deep learning ?
Sure does!