Hello Aarohi Your channel is very knowledgeable & helpful for all Artificial Intelligence/ Data Scientist Professionals. Stay blessed & keep sharing such a good content.
OSError: [WinError 126] The specified module could not be found. Error loading "C:\Users\patro\anaconda3\lib\site-packages\torch\lib\fbgemm.dll" or one of its dependencies. please reply me
Is there any way I can save the model's output?... like, if the number plate's 'ABC-123' or something can have it stored in a variable and display it within Python, because I need to store these number plate characters as a string in a dataFrame.
"Error executing job with overrides:" and AttributeError: 'Annotator' object has no attribute 'draw' these errors comes when tried to run this for second image.for the first image the number came.please hepl if anyone knows the bug earlier.
I explained only the part which I added and didn't explained which is already there. But those are just different functions inside a class to handle the input to model and then post processing and writing results.
I always have problem with OCR... I tried 3 different methods... Every time the region of number plate will be properly detected but then when I apply OCR on the cropped images... It just extracts #&((":! +3'srhed'hjhdv₹₹+:"+ This has given me better results than others but still OCR is detecting only about 50 percent characters correct... Idk what might be the issue
Hey, but for a real use case you would need a fast and reliable code. So for the real world C would be better right? Thanks for the video and detailed explanations!!
Thank you for your comment! I appreciate your perspective. In the context of my project, I chose YOLOv8 and EasyOCR for their ease of use and quick implementation, which was suitable for the scope of this demonstration. C can indeed offer better performance in terms of speed and resource efficiency, and it's a great choice for many applications, especially those where low-level control is essential. However, the trade-off is often development time and code readability. In a real-world scenario, the choice of tools and languages would depend on the specific requirements of the application, the available resources, and the balance between development time and performance.
I tried your code for my project but it didn't work well with my test video. It didn't read the plates properly. Do you have any ideas for this problem? Thank you.
You can try these: 1- Ensure that the training data is diverse, covering various plate sizes, fonts, colors, and conditions (like dirt or damage). 2- Include images from different angles, lighting conditions, and distances similar to your test video scenarios. 3- After detecting the plate region, ensure the image segment provided to EasyOCR is clear and well-aligned. 4- Experiment with different OCR settings
Proving every minute details in such a short video is really amazing...
Thank you so much
Hello Aarohi
Your channel is very knowledgeable & helpful for all Artificial Intelligence/ Data Scientist Professionals. Stay blessed & keep sharing such a good content.
Really amazing video. Please keep sharing such a knowledgeable content.
Thank you, I will
Great video, thanks for the lesson.
Glad you liked it!
Mam, Can you please tell me how to use webcam for real time input feed for this project
Really Amazing video mam, Thank you so much...
Glad my video is helpful :)
OSError: [WinError 126] The specified module could not be found. Error loading "C:\Users\patro\anaconda3\lib\site-packages\torch\lib\fbgemm.dll" or one of its dependencies. please reply me
Hey arohi, great video, I want to count distinct number plates and store them in array and return the count, what changes in code I should perform?
Is there any way I can save the model's output?... like, if the number plate's 'ABC-123' or something can have it stored in a variable and display it within Python, because I need to store these number plate characters as a string in a dataFrame.
Hey Aarohi suppose i want to return text in another module(.py fiile) how i can do that
Awesome video
Thanks!
Very nice
Thanks
great video , can you please tell , how i can change the location where the predicted video file is saved, im getting error in this. thx
"Error executing job with overrides:" and
AttributeError: 'Annotator' object has no attribute 'draw'
these errors comes when tried to run this for second image.for the first image the number came.please hepl if anyone knows the bug earlier.
beautiful
Thank you! 😊
Thank u ma'am,
I've been waiting for long time for this type of video..
But here u don't explained the predict_modify.py whole code..
I explained only the part which I added and didn't explained which is already there. But those are just different functions inside a class to handle the input to model and then post processing and writing results.
Hello mam can u provide a feature that can help to changes the input videos without Change the path name is the code.
You can pass the arguments through command line.
Thank
Is this code capable of recognizing Persian and Arabic license plates? And how can I change it to read these license plates?
Yes, You can. You need to change the language to Persian like this: reader = easyocr.Reader(['fa'])
I always have problem with OCR... I tried 3 different methods... Every time the region of number plate will be properly detected but then when I apply OCR on the cropped images... It just extracts #&((":! +3'srhed'hjhdv₹₹+:"+
This has given me better results than others but still OCR is detecting only about 50 percent characters correct... Idk what might be the issue
did you solve it
Hey, but for a real use case you would need a fast and reliable code. So for the real world C would be better right? Thanks for the video and detailed explanations!!
Thank you for your comment! I appreciate your perspective. In the context of my project, I chose YOLOv8 and EasyOCR for their ease of use and quick implementation, which was suitable for the scope of this demonstration.
C can indeed offer better performance in terms of speed and resource efficiency, and it's a great choice for many applications, especially those where low-level control is essential. However, the trade-off is often development time and code readability.
In a real-world scenario, the choice of tools and languages would depend on the specific requirements of the application, the available resources, and the balance between development time and performance.
How to make it as real world app? Can you give me an example please? For example c# gui can be used?
hello,please im having an error that says "FileNotFoundError: [WinError 2] The system cannot find the file specified" please can you help me
Which file not found? Share the entire error message.
I tried your code for my project but it didn't work well with my test video. It didn't read the plates properly. Do you have any ideas for this problem?
Thank you.
You can try these:
1- Ensure that the training data is diverse, covering various plate sizes, fonts, colors, and conditions (like dirt or damage).
2- Include images from different angles, lighting conditions, and distances similar to your test video scenarios.
3- After detecting the plate region, ensure the image segment provided to EasyOCR is clear and well-aligned.
4- Experiment with different OCR settings
Do you know any EasyOCR vs KerasOCR accuracy comparance? Any paper? Nice video btw.
I'm not aware of any specific paper.
can you provide me the link to your test video mam
hello
can you please add the test video file?
Great Video. How can we create file for running it on Android ?
You need a tflite model to run it in android app
Hi, can i get the code and datasets used in this project please ?
Mail me at aarohisingla1987@gmail.com
Running this project on kaggle/google colab
Great choice! Kaggle and Google Colab are excellent platforms for running data science projects.