Yours is such an underrated channel . The way you teach complex data analysis strats with patience and hands on approach is simply amazing. Thank you 😄
Hi, really clear explanation! I was looking into AI learning to analyse a othomasic picture from green houses for cracked and broken windows. As I was looking at your video it would be best to break down the mosaic in smaller tiles, train the model, add labels to the image and stitch the mosaic back together. What would be the best/maximum pixel image size for this method?
Thank you for the great efforts done here, I noticed that when I run the code for the same size of training I got different results each time, if there is any clarification for this point
First of all I apologize, I am using the google translator, I have a question, if I have a data set that I want to use to test this algorithm, how could I do it? Greetings from Chile friend
I'm trying to run the code with a sample image using the predict. Model () but for some reason I got an error message. Can someone advise here? Thanks in advance!
how could you detect the depth of the cracks? I know that through image segmentation you can get good results, but after that how to make measurments ? Any idea :s
Well if you want to predict the depth, you would need labeled data for the depth of each training image. This could be in the form of a numeric value and we could then use a regression model to approximate it.
Thank you for such great video... Can you explain how if I want to use that CNN on android studio programming so I can build an android app with that...? :)
Hello, thank you for this amazing tutorial. I am a student and currently doing a project for concrete crack detection but using mask r-cnn. I have only a little knowledge about cnns. Is there some way that I can use your code for the mask r-cnn? From what I see its a bit different and its confusing. Idk pls help me
Just flow the images in through the test generator we created in the video! new_images = test_generator.flow_from_dataframe( dataframe=, ... ... ) And get the results: model.predict(new_images)
Yours is such an underrated channel . The way you teach complex data analysis strats with patience and hands on approach is simply amazing. Thank you 😄
just stumbled on your channels and am glad i did , your passion is very infectious and your work is very admirable, well done
Thank you so much Ayenco!
Your consistency is admirable 💯💯💯💯👍
Thanks Shyam!
Hi, really clear explanation! I was looking into AI learning to analyse a othomasic picture from green houses for cracked and broken windows. As I was looking at your video it would be best to break down the mosaic in smaller tiles, train the model, add labels to the image and stitch the mosaic back together. What would be the best/maximum pixel image size for this method?
Thank you so much this saved me. btw if we run this with real images can we get accurate info?
Thank you. Saved my day
Thank you Sir! It helped me!
Thx for helping i wanted to ask do u have a tutorial that give us the width and length of the crack Positive Pictures
Thank you for the great efforts done here, I noticed that when I run the code for the same size of training I got different results each time, if there is any clarification for this point
permission to learn sir.thank you
For this project do we need high end computer with high processor Gpu ram?
First of all I apologize, I am using the google translator, I have a question, if I have a data set that I want to use to test this algorithm, how could I do it? Greetings from Chile friend
great content!
hey please make a video on NEU surface defect detection dataset
Thank U Sir. I was searching a work on curve lane detection for self driving cars. If you don't mind you may offer us♥️
What if I want to feed my own dataset as the test dataset? What should be the commands to detecting the cracks in those images?
Hi there!
How can I fix attribute errors? and can tensorflow work with intel graphics cards? if not, then is there anyway?
Hi, wanna ask after doing the detection how to do the test of training is crack image and show the confidence level on the each test image
I'm trying to run the code with a sample image using the predict. Model () but for some reason I got an error message. Can someone advise here? Thanks in advance!
how could you detect the depth of the cracks? I know that through image segmentation you can get good results, but after that how to make measurments ? Any idea :s
Well if you want to predict the depth, you would need labeled data for the depth of each training image.
This could be in the form of a numeric value and we could then use a regression model to approximate it.
how do you know that the positive string is 1 in binary and the negative string is 0?
Is there a possible to check on metallic surfaces? Thanks
Hi! Great channel. I'm coming from a Mechanical Engineering background and interested in learning DL in SHM. Any tips you can give me? Thanks
Any idea how you would liked to apply deep learning to SHM?
Thank you for such great video... Can you explain how if I want to use that CNN on android studio programming so I can build an android app with that...? :)
Hello, thank you for this amazing tutorial. I am a student and currently doing a project for concrete crack detection but using mask r-cnn. I have only a little knowledge about cnns. Is there some way that I can use your code for the mask r-cnn? From what I see its a bit different and its confusing. Idk pls help me
I am a student and I doing same your project for crack detection I you want to work with each other thats make me happy
Any updates on your project?
In case if i want to give single image and check whether it is positive or negative how am i suppose to do it ?
Hi Jyothi,
After the model is trained, you can call model.predict(image) to get the prediction for a single image.
How can i test few other cracked images for cracked/non-cracked prediction? thanks a lot
Just flow the images in through the test generator we created in the video!
new_images = test_generator.flow_from_dataframe(
dataframe=,
...
...
)
And get the results:
model.predict(new_images)