Kiana-Jahanshid / Age-Prediction

Age Prediction using CNN Regression and UTKFace dataset + Transfer Learning + wandb

Age Prediction 👶🏻🧒🏻👨🏻👵🏻

Description :

1_CNN Regression :

In this project , we want to predict people's age . Despite of previous projects which we used classification method using CNN , Now we are going to have one output which is an age . so here we have to use Regression method .

2_ Kaggle to Colab :

Here , we want to use a Kaggle Dataset named UTKFace . Now we can access dataset through kaggle , without downloading it directly : frist of all we have to create New API Token , It will download kaggle.json file on your google colab . here is the instruction :

1_ making a directory named Kaggle :

!mkdir ~/.kaggle 

2_ then copy kaggle.json file into that :

!cp kaggle.json ~/.kaggle/ 

3_ using below command our dataset will be download :

!kaggle datasets download -d jangedoo/utkface-new 

4_ now we need to unzip and use it :

!unzip -qq utkface-new.zip 

-------------------------------------------------------------------------------------------------------------------------

3_ Using wandb :

wandb or Weights & Biases, is a machine learning platform that allows you to track & visualize different aspects of the model training process in real-time.
here is the project wandb link :
https://wandb.ai/kianajahanshid/Age_prediction_UTKFace?workspace=user-kianajahanshid

How to install

Run this command to install used libraries :

pip install -r requirements.txt 

How to run

you can run Age_predictor_UTKFaceDataset.ipynb file and train model , or else you can load_model with .keras file which is in below google drive link :

https://drive.google.com/file/d/1-0K9oUBkDRGl8onp2ABVOZY2AYF5Y3ok/view?usp=drive_link

Results :

• with our own picture :

• test other people :