Skip to content

ostjen/cv_task

Repository files navigation

The computer vision task is to create a neural network which takes an image of a face as input and returns its orientation - upright, rotated_left, rotated_right, or upside_down - and use this neural network to classify and correct the images in the test set.

Dependencies

OpenCV

Pandas

Numpy

Scit-learn

Keras

Model:

Usage

Unzip all images into the root of the repo in a 'train' directory

Train

$ python pure_convolutional.py

Rotate images and generate numpy array file(.npy)

$ python change_images.py

Results

in 20 epochs the model reached almost 95% accuracy

Alt text

check the demo :)

About

Computer vision task using CNN's

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages