ardamavi/dog-cat-classifier

Dog and cat image classifier with deep learning

Dog-Cat Classifier

By Arda Mavi

Dog and cat image classifier with deep learning.

Example:

Dog: 0.92035621 Cat: 0.04618423	Cat: 0.90135497 Dog: 0.09642436

Layer outputs of test photographs:

Layer: 1 Kernel: 4	Layer: 2 Kernel: 16	Layer: 3 Kernel: 10

Look up Data/Layer_Outputs folder for other outputs.

Using Predict Command:

python3 predict.py <ImageFileName>

Model Training:

python3 train.py

Using TensorBoard:

tensorboard --logdir=Data/Checkpoints/./logs

Model Architecture:

• Input Data Shape: 64x64x3

Layer 1:

• Convolutional Layer 32 filter Filter shape: 3x3

• Activation Function: ReLu

• Max Pooling Pool shape: 2x2

Layer 2:

• Convolutional Layer 32 filter Filter shape: 3x3

• Activation Function: ReLu

• Max Pooling Pool shape: 2x2

Layer 3:

• Convolutional Layer 64 filter Filter shape: 3x3

• Activation Function: ReLu

• Max Pooling Pool shape: 2x2

Classification:

• Flatten

• Dense Size: 64

• Activation Function: ReLu

• Dropout Rate: 0.5

• Dense Size: 2

• Activation Function: Sigmoid

Optimizer: Adadelta

Loss: Binary Crossentropy

Adding new train dataset:

If you want to add new dataset to datasets, you create a directory and rename what you want to add category (like 'cat' or 'phone').

If you want to add a new training image to previously category datasets, you add a image to about category directory and if you have npy files in Data folder delete npy_train_data folder.

Note: We work on 64x64 image also if you use bigger or smaller, program will automatically return to 64x64.

Important Notes:

• Used Python Version: 3.6.0
• Install necessary modules with sudo pip3 install -r requirements.txt command.

dog-cat-classifier's Issues

Any JPG returns "It is a dog !"

In passing test_dog.jpg or test_cat.jpg, both return "Its a dog!". Maybe a problem on my end? Im running train.py first and then predict.py

thanks

Some questions.

How could you show the predict value like "Example"?

How can I make my own npy train data?

First of all thank you for your amazing work, I'm planning to implement ur code using my own data, im wondering how can i make my own npy train data?

run the classifier on tensorflow GPU

Hi,
Do you know if I can easily change your code to make it work on GPU ?

kullanılan mekanizmayı readme de yazabilir misiniz?

train.py problem

patrick@123:~/Desktop/Dog-Cat-Classifier-master$ python predict.py test_cat.jpg Using TensorFlow backend. 2019-10-25 00:19:23.323125: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 2019-10-25 00:19:23.340188: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 1992000000 Hz 2019-10-25 00:19:23.340916: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x694d4f0 executing computations on platform Host. Devices: 2019-10-25 00:19:23.340974: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (0): Host, Default Version [[1. 0.]] [0] It is a cat !

I have entered train.py
without removing or adding any image
but accurancy prints [[1. 0.]]
could you tell me why?
thanks.

How to Make detection of dog and cat along with classification?

This project does classification very well but how we can do detection also inside the image.

Needs Update...

some libraries use outdated methods...

get_dataset.py needs to be corrected as:

at line 14
img = imresize(img, (img_size, img_size, 3))
must changed as
img = Image.fromarray(img).resize(size=(img_size, img_size))

add line 7:
from tensorflow.keras.utils import to_categorical
change line 47 to:
Y = to_categorical(Y)
must be added

How to detect the dog-cat image inside the image?

The project here has done classification very well but can we something which will detect the cat/dog in the image also.