NahidEbrahimian/Deep-Learning-Mini-Projects-Tensorflow

Repository for Deep Learning Mini-Project with Tensorflow

Deep-Learning-Course

01- CNN and MLP

• Test accuracy comparison between MLP and Deep Learning

Dataset	MLP (Machine Learning)	CNN + MLP (Deep Learning)
Mnist	0.97	0.98
Fashion Mnist	0.86	0.89
Cfar 10	0.37	0.63
Cfar 100	0.13	0.28

02- Sheikh-Recognition

• Train Neural Network for normal person and sheikh images classification

• train.ipynb

• inference.py

Dataset:

contain images from two classes, normal person and sheikh

For inference run the following command:

!python3 inference.py --input_image test/image3.jpg

03- Sheikh-Recognition-bot

• Train Neural Network for normal person and sheikh images classification

• train.ipynb

• bot.py

Dataset:

contain images from two classes, normal person and sheikh

Usage:

1. Click here to open the chat with the bot in the Telegram app

2. Start the bot and send him a photo

04- Persian Face Recognition

• train.ipynb

• preprocess.py

• inference.py

Preprocessing

Preprocess stage consists of 4 common stages: detect, align, represent and verify. link: Github

1. You must first install retinaface:

!pip install retina-face

2. Run the following command to apply preprocessing:

!python3 preprocess.py --input_images_dir "./input_images" --output_dir "./output_dir"

05- 17Flowers Classification

Dataset:

Contain images from 17 classes of flowers in two subset, train and test.

Dataset link: Flowers

Result:

Comparison accuracy of pretrained models that used in transfer learning on test data:

Model	Accuracy
Vgg16	0.67
Vgg19	0.70
ResNet50V2	0.82
MobileNetV2	0.37

07- UTKFace-Age prediction-Regression

• Train Neural Network on UTKFace dataset using tensorflow and keras

• train.ipynb

• inference.py

Dataset:

Dataset link: UTKFace-dataset

inference:

1- First install retina-face

!pip install retina-face

2- Run the following command:

!python3 inference.py --image_path 'input/08.jpg'

08- Gender Recognition

• Train Neural Network on gender-recognition dataset using tensorflow and keras

• train.ipynb

• inference.py

Dataset:

Dataset link: celeba-dataset

10- Houses Price Image Regression

• Train Neural Network for house price prediction using images

• train.ipynb

• inference.py

Reference: KerasRegressionandCNNs

11- DCGAN_on_Mnist_dataset

• Training DCGAN on Mnist dataset

Reference: tensorflow-tutorials

12- DCGAN_on_Celeb_A_dataset

• Face Generator, Training DCGAN on celeba dataset

Dataset link: celeba-dataset