kaggle_Camera_Model_Identification

This repo is used to record the code and files that I writed when kaggle-Camera_Model_Identification.

---

The code includes downloading and augmenting the data, caffe model and super parameters, ensembling the models and testing models.

First step:

Augment the data using the eight possible processing operations.

Second step:

Trained the models (Resnet50, Resnet101, InceptionResnetV2, SeResneXt50) using the pretrained models.

Third step:

Test the models using single crop or multi crops.

Last step:

Ensemble the models.