This is a solution of the problem in the datafountain that "China Meteorological big data algorithm and application contest -
algorithm competition".
The main idea to solve the task is How to train a better classification for the given data.
Here, we try serverl models separately to learn the specific classifier. Including the VGG and Inception model.(inception, xception,
densenet model can also be achieved easily through just modifying model.py file). There isn't any other operation skills applied to the
solution. So, if you combine more image processing, mayhe you are able to get a more accurate and robust result.
The project mainly contains the following files:
dataset.py process the given data
train.py train the model
model.py build the network through transfer learning
test.py test and output the results
if you use the project, you firstly need to put the given image into corresponding folders.