Deep Learning- Handwritten Digits Recognition Tutorial | Tensorflow | CNN | for beginners

[@gustavofelicidade_]

I did a test to get a job using this model as base of my technical challenge, many thanks!

[@kshitijvarshney]

The model.add(Conv2d()) doesn't works the same way anymore. the activation function is now integrated as an argument within the function along with a kernel initializer. My code didn't work but after a little finding this worked: model.add(Conv2D(64, (3, 3), activation='relu', kernel_initializer='he_uniform', input_shape=x_trainr.shape[1:])) #convolutional layer

[@aparnasahu2835]

Same problem but remain not working

[@aparnasahu2835]

Please can you help me

[@Kaustubh56]

Thanks! This was really helpful! Could you please do a video on generating Transcription of a speech or an interview in English? Saw a couple of such videos but those were clumsy or didn't work.

[@gowthamireddykovvuri]

Thanks a lot sir...❤️ from bottom of my heart ! This video helped me to complete my Project 🤗

[@shwetanshusood9450]

Smartknower waale😂😂

[@rohitpaswan0001]

@Gowthami Reddy Kovvuri can you tell me how to do because i also got this same project

[@samr405]

You're a lifesaver. Thank you.

[@AbdulREssani34cP]

Very good video, easy to understand. Helped me a ton with my class project

[@subhradeepmaji4925]

@DeepLearning_by_PhDScholar It's really a good tutorial. It will help me in my university project. Thank you for your tine to prepare such kind of video.

[@yervandyahyeyan5235]

Thanks for the best tutorial in the world, I have a question, Can you send last video Cell?

[@shraddhasumandash5465]

Awesome explanation sir . Thankyou for the video

[@LakshayBharadwajBIT]

I am having validation accuracy of something 0.1 and loss of about 10 whats wrong and even its training itself on 1313 while fitting the model idk why whats happening can someone suggest something. It sould train itself on 42000 sets but instead it is doing that on 1313 sets. For refrence in video at 36:45 a little before that.

[@Lyff]

Same. Did you find any solution to that?

[@aliengineroglu8875]

Thank you for your efforts. Did you share the code? I couldn't find it?

[@simonlaplace9790]

there is an error (TypeError: Image data of dtype object cannot be converted to float) when I tried to plot the paint image with plt.imshow(img)

[@p.kiranmai2466]

same.... did u find the solution??!!

[@DD27_27]

Hi sir, Thanks for your video. Sir I am working on spintronics based neuron and synapse so I have kind of my own Neuron activation function equation and synaptic variation value. I want to test the MNSIT dataset to obtain the accuracy for such hardware models. Can you suggest some possible solution on how to add those as neurons and synapse at the last dense network stages.

[@techinfo89]

Your explanation was amazing, sir, thank you so much

[@cheems6559]

Sir, how do u decide the number of layers....is it hit and trial or there is some method behind this

[@tanishqrastogi1011]

well its been 3 years so i hope you have the answer to this video. can you please tell me how to choose number of layers and units per layer from your experience?

[@gustavofelicidade_]

Thank you from Brazil!!!!

[@roopsagarteella]

Learned a lot sir, thank u.

[@gueg2298]

thanks bro it takes for 10 epoch to work correctly

[@narendramanchayala9349]

Bro At last you saved the images of numbers through paint Is there any pattern like where to save this bcoz I’m getting type error after saving and giving them

[@jonas1351]

have you found a solution?

[@minhhai3559]

Thank you so much, can you give me your code... i get error at step training model

[@ramamunireddyyanamala973]

Thankyou Sir , your explanation is very good

[@ampreagent]

Great explanation!

[@abdimoussa4220]

Hello and thanks for your videos Because it's helps me a lot and I would like that you make us a video with speech-recognition thanks also

[@shubhampatel4426]

super cool and very informative, love from india : )

[@oxydol3456]

it's clear and easy to follow thanks

[@jahanvi9429]

Thank you very much! It was of immense help

[@milindsoorya5870]

at 24.05 why did you resize the image? Why an extra 1 dimension?

[@yogeshdange7210]

How we can use Synaptic weights of device for MNIST image Reconization ? Is any code avalible ?

[@sam_oyeleye]

it would be great if you talked a little bit slower, sometimes it's hard to hear the words you pronounce because it is said a bit faster than necessary...Good video nontheless

[@aaaqaaaa2720]

Hi sir I try to excute this code and i have problem that sequential is not defind why ? Thanks advance

[@kk0511]

Is the demo video commands necessary on from 50:00

[@lahariyenugu6247]

Why it is showing pool_size' is not defined can anyone please reply?

[@nadergamalrihan9054]

Thank you, very helpful

[@manmeetsingh8349]

hello sir, can you share the ipynb file of this project, that will be really helpful.Thanks

[@shwetanshusood9450]

Really interesting ☺❤

[@RaviSankar-ln3ki]

Thank you.

[@smegala3815]

Thank you sir

[@kunalkhatri2720]

It's Training on 1313 samples only, whats the reason behind that

[@adityamittal7335]

sir please help I want to implement this code with EMNIST data set

[@MightyFrostDragon]

Very pedagogical, thank you so much. It was helpful

[@chyrilnarvasa2513]

sir can this system recognize more than 1 digits per image??????

[@shikhamishra4962]

sir please tell me how to resolve this error . I have to show this project in my college.Sir please reply

[@chemophile14]

Very nice and quiet interesting

[@_mohammed_kaif_ahmed_]

tq sir

[@muhammadzain-ul-abideen3228]

When my model starts training, it only train 1313 samples in each epoch.... Where am i wrong? Plz Help

[@joshuagajjala176]

Even i too got the same 1313 samples , maybe we need to check with validation I guess

[@muhammadzain-ul-abideen3228]

@@joshuagajjala176 I've found a solution, just set a parameter batch_size = 1 in model.fit().

[@joshuagajjala176]

@@muhammadzain-ul-abideen3228 Thanks for helping this beginner . If you would like to share and work together please ping me in mail - [e-mail]

[@ekanshmishra4517]

@@muhammadzain-ul-abideen3228 thanks man🤝it was really a help. It will be glad to work with you

[@samr405]

​@@muhammadzain-ul-abideen3228 Thank you. I was having the same problem and I never would have figured this out.

[@pradixrajpoot]

I tried handwritten digit recogniser which recognizes that the image is of which number,but after training the model it is predicting wrong number.I have saved the image from google images. My friend told me to preprocess the image as that of mnist dataset format.But I didn't got what he wanted to say. Could you help me please along with code requirement.

[@ekanshmishra4517]

Yes i am also facing this same problem.can you help me in this on how to improve the accuracy of handwritten images?

[@aaaqaaaa2720]

Can any one help me when i built the neural network i have sequential is not defined

[@沛澐林]

thank you!!!

[@naimatulmd3306]

Thanks...❤️❤️❤️

[@shikhamishra4962]

error: OpenCV(4.1.2) /io/opencv/modules/imgproc/src/color.cpp:182: error: (-215:Assertion failed) !_src.empty() in function 'cvtColor'

[@deeplearning_by_phdscholar6925]

Please share the lines of code where you get error. It seems like you are applying colorspace to a gray level image. If your image is already gray image i.e its bit depth is 8 you dont need to apply colorspace to show it. But. I can't say anything unless you share lines of code you got problem

[@shikhamishra4962]

@@deeplearning_by_phdscholar6925 gray = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB) sir i am running this code on google colab

[@shikhamishra4962]

sir above line is showing error

[@BabarAli-mq6wk]

Sir you are good

[@Scholars-ct1pr]

Hi....woul you be able to provide the ypinb file?....Cheers

[@soumilyade1057]

The viewer must be familiar with the concepts before watching the video...

[@dianathussain2056]

I hope my text will find you in good health I saw one of your video on KZbin about digit Recognition using CNN I need to discuss some query if you don't mind Actually it's my university project I need a system a proper webApp having options of upload image , or Android App having options of capture an image after that it recognises the digits in the image, I will use the Mnist dataset for this project I need some help of yours please Will be waiting for your response Thank you

[@dianathussain2056]

Hi Can you share the source code please?

[@shivamdubey4783]

can you upload for video also

[@beck02ify]

Freaking nice , btw What are the 42000 in the training phase ?

[@deeplearning_by_phdscholar6925]

Thank you so much for your interest and glad to hear that you like the video. Sorry for being late in replying. during training model.fit(x,y, validation_split= 0.3) so Training data was 60,000, and test was 10,000 So in order to further split training into validation data, I defined validation_split= 0.3, 60,000 * 0.3 = 180000 so 18000 => validation dataset 60,000-18000 = 42000 (Training dataset after subtracting validation dataset) you can change validation_split= 0.3, to any percentage you want, but its usual practice to divide into 70 vs 30 percentage I hope I have answered your question. Stay blessed, keep supporting my channel .

[@beck02ify]

@@deeplearning_by_phdscholar6925 Thx for the answer !*

[@anujshaw3570]

@@deeplearning_by_phdscholar6925 thanks for this explanation :)

[@shivamdubey4783]

sir for every prediction it is giving 1

[@shivamdubey4783]

sir giving me wrong prediction for 7 it is giving 1

[@dhyanadave5409]

Hello sir, I got an error while creating a deep neural network, when I try to add a third convolution layer it shows me an error. Sir, I am doing my project, so please help me to solve it asap. Error: InvalidArgumentError: Negative dimension size caused by subtracting 2 from 1 for '{{node max_pooling2d_2/MaxPool}} = MaxPool[T=DT_FLOAT, data_format="NHWC", explicit_paddings=[], ksize=[1, 2, 2, 1], padding="VALID", strides=[1, 2, 2, 1]](Placeholder)' with input shapes: [?,1,1,64].

[@aakmad18m36]

sir can i get your record.mp4 video through drive link pls ....helpp

[@agnivesharya4462]

heyy i am getting an error----error: (-215:Assertion failed) !_src.empty() in function 'cv::cvtColor' what to do?

[@nurly5209]

have the same issue, did you find solution?

[@BigyanChap]

👍

[@monicagunjal2977]

Code link please...

[@syedfarooqshah5984]

Bro i need code

[@shivamdubey4783]

can you plllzz atleast reply

[@shikhamishra4962]

sir please reply

[@nemo-fx2ki]

i dont know why when i type plt.imshow(img) it is not showing the image even i wrote this to show the path it is still not showing import matplotlib.pyplot as plt img_path = r'C:\Users ere\Documents\eight' img = plt.imread(img_path) plt.imshow(img) plt.show()

[@deeplearning_by_phdscholar6925]

Thank you for you're concern. Your name of image must contains image extensions as well such as at the end it must be eight.jpg or eight.png. please add complete image name along with extension and try again. Thank you

[@aakmad18m36]

sir can i get your record.mp4 video through drive link pls ....helpp

[@aakmad18m36]

sir can i get your record.mp4 video through drive link pls ....helpp