Improve validation accuracy
5:46
4 жыл бұрын
COVID-19 | Screening test metrics
12:56
How to freeze a TensorFlow Model
6:43
How Apple can Fix FACE ID
8:20
6 жыл бұрын
Hello World TensorFlow-Android
27:10
7 жыл бұрын
MNIST in TensorFLow  In-depth analysis
35:52
Lecture 8:Basics of Neural Networks
24:27
Lecture 4: Basics of TensorFlow
13:37
Lecture 2: Setting Up IDE and VM
8:54
Lecture1: Introduction
4:35
8 жыл бұрын
Пікірлер
@mrnerdy3501
@mrnerdy3501 6 ай бұрын
Why do you say that 3x3 and 5x5 convolutional filters will leave the number of channels untouched? I can choose any number of output channels for any NxN convolutional filter, because I can choose any n_out NxN convolutional filters to slide over my input.
@souravnair7828
@souravnair7828 Жыл бұрын
Hey. Is there any way to lock the frozen model and the model can only be unfreezed from my end
@MIKELAURENCEFERNANDEZ
@MIKELAURENCEFERNANDEZ Жыл бұрын
hi sir hope you can help me,, how to create unclassified image if that certain image is the object in dataset?
@carlosroquesuarezgurruchag8681
@carlosroquesuarezgurruchag8681 2 жыл бұрын
Thx!
@telugunewtons2304
@telugunewtons2304 2 жыл бұрын
How is the accuracy of the validation set affected if we increase the percentage of validation set? What happens when we reduce it?
@dharmendrakumar-ws3fk
@dharmendrakumar-ws3fk 2 жыл бұрын
hi sir plese rovide me doc and code with model summary for alexnet and resnet ..pls
@dharmendrakumar-ws3fk
@dharmendrakumar-ws3fk 2 жыл бұрын
hi sir good morning please provide me the doc and code with a model summary for alexnet and Resnet ..pls
@alexiscarlier2773
@alexiscarlier2773 2 жыл бұрын
Hi Thank you. Do you the behavior of loss functions for multioutputs regression??
@mtariquegauri
@mtariquegauri 3 жыл бұрын
Thank you. 🙌🏻
@Roo5331
@Roo5331 3 жыл бұрын
Hi, I have tf model .pb file, but how do I find my input and output node?
@krishnachauhan2850
@krishnachauhan2850 3 жыл бұрын
Of course data augmention combat with the problem of overfitting almost always but you can't do it in research
@mrunalinim8020
@mrunalinim8020 3 жыл бұрын
How to find out the names of i/p and o/p nodes?
@nitesharu1269
@nitesharu1269 3 жыл бұрын
i have made a model and its file is in pkl format, so what steps to follow to convert it in tflite format
@aadiegoaa96
@aadiegoaa96 3 жыл бұрын
Thanks for the video from Mexico
@aadiegoaa96
@aadiegoaa96 3 жыл бұрын
Thanks for the video from Mexico
@yzhang4970
@yzhang4970 3 жыл бұрын
very helpful video
@austinvest9858
@austinvest9858 3 жыл бұрын
again great concise intro into many complex deep learning validation concepts. thank you.
@austinvest9858
@austinvest9858 3 жыл бұрын
This video is closely relatable to www.mathworks.com/help/deeplearning/ug/train-deep-learning-network-to-classify-new-images.html a mathworks Matlab example. This video is very concise which very helpful got getting that big picture or transfer learning. Thank you.
@meriljayasinghe9292
@meriljayasinghe9292 3 жыл бұрын
What is the AI technique used here? CNN?
@chitishagunnoo2134
@chitishagunnoo2134 3 жыл бұрын
Genius👏👏👏
@wliiliammitiku3996
@wliiliammitiku3996 3 жыл бұрын
how to apply data augumentation
@jatayubaxi1736
@jatayubaxi1736 3 жыл бұрын
Nice.
@Gopikha6
@Gopikha6 4 жыл бұрын
Lecture is very clear and more understandable.Is it possible to predicts colors in the RGB image using deep learning?
@learndlcodetf5331
@learndlcodetf5331 3 жыл бұрын
Thank you very much. Glad you like it. I have to think about it. But talking right off my mind, an RGB image is made up of pixels. Each pixel has three integer values (r,g,b). Using regression you can predict a number and round off to nearest integer. But out of curiosity why do you need deep learning to do this task? when you have the image, i.e the input tensor, you already know the colours of an image right? if you want to mimic an image you can try auto encoders.
@Gopikha6
@Gopikha6 3 жыл бұрын
@@learndlcodetf5331 Thanks for your reply. Let me try it.
@lakshaniekanayake3592
@lakshaniekanayake3592 4 жыл бұрын
WoW Great Tutorial .All in One. sir How to get count of leaves of a small plant?
@tianmingguo8271
@tianmingguo8271 4 жыл бұрын
Really great explanation with details. Thank you.
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
Thank you very much Tim :)
@travelWithSamriddhi
@travelWithSamriddhi 4 жыл бұрын
hey can you help me please.....why am i getting error "IndexError: tuple index out of range" while Freezing the graph?????
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
which version of TF are you using? could you give a little more details about your tensorflow model?
@travelWithSamriddhi
@travelWithSamriddhi 4 жыл бұрын
@@learndlcodetf5331 2.3 Which version should I use?
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
@@travelWithSamriddhi TF 2.0 and above made lot of changes. Actually might not even need to freeze. In this example, I think I used 1.13.1 What is your main purpose of freezing? if if is to convert model to TFLite there are more direct ways in TF 2.3
@travelWithSamriddhi
@travelWithSamriddhi 4 жыл бұрын
@@learndlcodetf5331 I created .tflite file without freezing when I applied it on Android the prediction was wrong! That's why I thought I should try this approach!
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
@@travelWithSamriddhi this has no relation with the accuracy. Try for things that might actually lead to mispredictions in Android. For example, did you do the same preprocessing steps that you did in Python?
@yaohongtao4763
@yaohongtao4763 4 жыл бұрын
Hi, thank you for your tutorial. Can I ask which TensorFlow version you used?
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
It is based on TensorFlow Version 1.x. TF moved the TFlite package from contrib in later versions. Which version of TF are you working with? I can help you find right package location
@sundararajan3306
@sundararajan3306 4 жыл бұрын
Hi vikraman, Thanks for this excellent video. As for cross entropy, another interpretation is that when predicted value is more sure ie. Higher probability(say, 0.9) the loss or penalty is lesser and vice versa.
@08ae6013
@08ae6013 4 жыл бұрын
is it similar to nn.AdaptiveAvgPool2d() in pytorch ? pytorch.org/docs/stable/nn.html?highlight=nn%20adaptiveavgpool2d#torch.nn.AdaptiveAvgPool2d
@anagrarian
@anagrarian 4 жыл бұрын
This is the latest (June 2020) solution for converting tensorflow models to tflite models. pip install tf-nightly -- do the magic. colab.research.google.com/gist/jvishnuvardhan/23066f1a722fc566b437a5210d1b97b0/ssd_saved_model_tflite_conversion.ipynb?authuser=2#scrollTo=2-STWtT7CQd8
@timdick5149
@timdick5149 4 жыл бұрын
Thank you very much for this awesome lecture! It made me understand the principle much better! Great job! Maybe as a tip, it would be nice to get some formulas for possible values of stride s in both the integral and fractional convolutions.
@erolgyte
@erolgyte 4 жыл бұрын
i changed tf vrsion to 1.15 but i got this error AttributeError: module 'tensorflow_core.contrib' has no attribute 'lite'
@MrAbehonest
@MrAbehonest 4 жыл бұрын
Same here. I'll let you know if I figure it out
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
Hi Erol, TensorFlow moved the location of "lite" function. Try "from tensorflow import lite" . Let me know what you get
@erolgyte
@erolgyte 4 жыл бұрын
@@learndlcodetf5331 same error. AttributeError: module 'tensorflow' has no attribute 'contrib'
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
The error is caused due to the location of TFlite package. Could you reconfirm tensorflow version by "print(tensorflow.__version__)"
@erolgyte
@erolgyte 4 жыл бұрын
@@learndlcodetf5331 1.15
@erolgyte
@erolgyte 4 жыл бұрын
AttributeError: module 'tensorflow' has no attribute 'contrib'
@sanjeev_mohan
@sanjeev_mohan 4 жыл бұрын
contrib was remove from Tensorflow 2.0 and above thus this error
@MrAbehonest
@MrAbehonest 4 жыл бұрын
Thanks so much for this video! You've absolutely made my day. I've been working on getting a frozen file properly made (with TF 1.x) for the better part of 2 days and your explanation was so concise that I finally got it to work!
@nurzhanamantay8729
@nurzhanamantay8729 4 жыл бұрын
Thank you 👍🏻
@amalnn8979
@amalnn8979 4 жыл бұрын
How did depth increase from 3 to 32?
@parmarsuraj99
@parmarsuraj99 4 жыл бұрын
That Thumbnail!!!🔥😅
@pulakammalathy6968
@pulakammalathy6968 4 жыл бұрын
After following the complete process, on running the app, its Showing app keeps stopping.in load library line of code, what should be the libname.
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
Hi, Google, updated the docs, new step by step tutorial next month
@Vathananable
@Vathananable 4 жыл бұрын
Sir how label a photo as unknown using tensorflow if the photo is not from 3 classes.
@ShantanuVishnoi
@ShantanuVishnoi 4 жыл бұрын
Very useful :) great job 👍
@rashmithareddy1220
@rashmithareddy1220 4 жыл бұрын
Thanks a lot sir. Your tutorial helped me a great deal in my project.
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
Hi Rashmitha, glad you like it. Google has updated settings for Android ML implementation. New video coming soon!
@nawalhamed6394
@nawalhamed6394 4 жыл бұрын
Hi, If I have .pb model file and I want to convert it to .h5 can I do that?
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
Hi Nawal, I suppose you want to use it with Keras API. In that case load the '.pb' file with TF API. Keras is a part of latest TF APIs. So after loading you can use tf.Keras functions to save and modify
@nawalhamed6394
@nawalhamed6394 4 жыл бұрын
@@learndlcodetf5331 OK, great. Thank you
@abdulazizmohamed7332
@abdulazizmohamed7332 4 жыл бұрын
how i can convert darkflow model (yolov2) to tensorflow model please??
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
Hi, check out this repo
@piumideshani3595
@piumideshani3595 4 жыл бұрын
I have a keras model how can I implement this model to this method please tell me
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
Hi, could you please elaborate?
@dandybramasta979
@dandybramasta979 4 жыл бұрын
Thank You for uploading this tutorial. It helps me a lot. By the way, I have some problems with my code. When I trained my model the validation accuracy keeps fluctuating, and my validation loss tends to increase, also when I tried to predict a sample image, my model keeps giving me the same output prediction. I have posted my question in stackoverflow too : stackoverflow.com/questions/61570271/python-cnn-image-classification-it-gives-me-always-the-same-prediction could you please help me with it? I appreciate it a lot Thank You
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
Hi there! your model has not learned properly. If you are getting the same output always there are two possible cases: 1) You have not shuffled the dataset. 2) Your dataset is highly imbalanced. How many images you have in each set?
@tosay1393
@tosay1393 4 жыл бұрын
Screen is too dull
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
Hi, sorry about that. I've posted the codes in case you want them. Also new tutorial with the latest API is on the way. This time, I'll pay extra attention to screen brightness. Good day!
@ioemiracles5682
@ioemiracles5682 4 жыл бұрын
Thank you for your tutorial. kindly tell si it ANN?
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
hello, yes it is very much a neural network. It is a fundamental(basic) neural network that contains neurons for input and output layers. I have not added hidden layers to make things simple.
@muhammadrasikhkamran3017
@muhammadrasikhkamran3017 4 жыл бұрын
@@learndlcodetf5331 kindly tell how many neurons it contains at input and output layers?
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
​@@muhammadrasikhkamran3017 hi, there are four neurons at the input layer corresponding to the four features used and 3 neurons at the output layer corresponding to the three classes. A single weight matrix of size(4,3) connects the input layer neurons to the output layer neurons.
@muhammadrasikhkamran3017
@muhammadrasikhkamran3017 4 жыл бұрын
@@learndlcodetf5331 we can also add hidden layer. Kindly guide code to add 2 hidden layers.
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
Hi, it is much easy with latest TF API and Keras by adding dense layers, like model.add(Dense(6)) after input layer. But if you want to follow my approach in TF 1 version, you can declare another weight variable of shape output of the first layer(input) and input of the final layer(output)
@vahit·keskin
@vahit·keskin 4 жыл бұрын
Please help me
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
Hi, how can I help you?
@vahit·keskin
@vahit·keskin 4 жыл бұрын
@@learndlcodetf5331 I am sorry my English level is a bit low. The snapshot of the camera I want to make. I have txt and tflite files printing the txt equivalent of the image taken from the camera.
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
hi@@vahit·keskin, could you please explain more? what do you mean by snapshot?
@vahit·keskin
@vahit·keskin 4 жыл бұрын
Hi @@learndlcodetf5331 The image from the camera. For example %99.8 probability class1 owned, %97.8 probability class2 owned and %98.8 probability class3 owned
@khanwaqar7703
@khanwaqar7703 4 жыл бұрын
Very cute and nice explanation, Thank you so much
@SS-ld5vb
@SS-ld5vb 4 жыл бұрын
Thank you very much sir but my app crashes at the movement it starts when i put .pb file i created from your git account by simply coping the code.But it works fine when i put your .pb file from git account.please help me to get out from this sir..
@learndlcodetf5331
@learndlcodetf5331 4 жыл бұрын
Hi, Google has updated the libraries. It mostly requires TFLite for Android apps. I'll make a step by step on Android soon
@shabnampathan8975
@shabnampathan8975 4 жыл бұрын
One of the best tutorial.I have trained tensorflow based GAN model.Now I want to make an Android app but I am totally new to it so I am confused on how to map the functionality into Android. I have a question there are two scripts train.py and test.py ,so do I have to map functionality of test.py into Android or train.py ?