Assignment 3 of Machine Learning Course: miniature of ImageNet classification challenge.
Achieved Accuracy of 83.30%, placed to 2nd rank.
Using Xception Model, loading pre-trained Imagenet weights and fine-tuning to achieve better validation accuracy.
10% of the data used for validation, rounding up to 90,000 train images, and 10,000 validation images.
Re-structured dataset to the following stucture:
data/
train/
class1/
class2/
validation/
class1/
class2/
Install dependencies before running the program
pip3 install -r requirements.txt (Python 3.5.2)