In-Class Competition: Facial Expression Classification
Automatic human facial expression classification has lots of potential real life applications, such as customer service systems, personalized advertisement, and entertainment. In our assignment 2, we will focus on building a deep learning model for
Facial Expression Classification
.
The training dataset and testing dataset are made available in the Data section. The training set consists of 16,175 examples for you to build your model. The test set consists of 3,965 examples which will be used to determine your score. Please check the Data section for more details.
Please note that the data are from a real Kaggle competition, https://www.kaggle.com/c/challenges-in-representation-learning-facial-expression-recognition-challenge/data
Specific Tasks:
1. Develop your Deep Learning models for facial expression classification using Keras.
2. Gain experience in training neural network models.
3. Follow the submission file format and instructions to submit your predictions in Kaggle.
4. Zip your code and your final prediction file, and submit the zipped file in our course blackboard.
Some Useful Resources:
[Building powerful image classification models using very little data](https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html)
[Available models in Keras](https://keras.io/applications/)
[Details about the network architecture VGG-16](https://gist.github.com/baraldilorenzo/07d7802847aaad0a35d3)
[Details about the network architecture VGG-19](https://gist.github.com/baraldilorenzo/8d096f48a1be4a2d660d)
[Scikit-image: an image processing tool in Python (You may use imread, resize, and more functions)](https://scikit-image.org/)
[OpenCV-Python Tutorials (Scikit-image is enough for this project, though)](https://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_tutorials.html)
[Convert a grayscale image to a 3-channel image (a simplest approach by duplication)](https://stackoverflow.com/questions/40119743/convert-a-grayscale-image-to-a-3-channel-image)