This project is created by Han X. Zhang and Zheng H. Xue in School of Artificial Intelligence, Nanjing University, which is part of the course project of AI Programming in 19Spring.
Baseline code is mostly based on Transfer_learning_tutorial by Pytorch.
The goal of the project is to teach the machine to do flower classification. There have been tons of researchs in flower classification, together with many large datasets. However, it's of great fun to start the project from scratch and make our own flower dataset, adding flowers that really fascinates us. Considering the constraint of resourses, we decide to base our project on transfer learning, and enrich the dataset if needed. The baseline algorithm, i.e. the most simple transfer learning via fine tuning, reaches an accuracy on nine flowers of 88.9%. Most importantly, the whole experience of setting up a project, though not advanced, is worth reconsidering.