As part of our final group project, students shall implement a neural network to gain a better understanding of machine learning in practice. To achieve our goal, we try to get the best possible accuracy. To fully understand this dataset and the classification problem, we provided a detailed description of task and its issues. For having a broader understanding of what research contributes to this task of image recognition, we introduce a summary of the latest publications and developments around the image classification field. For our implementation we decided to use a VGG16 model architecture. We go into detail about our model architecture and explain all of its components and corresponding layers. In the end, we analyse the results of our training process and compare it with similar implementation and the state-of-the-art models.
Clone the repository, to get our Notebooks and Project Report.
git clone https://github.com/bohniti/cifar-10.git
- You need git to get the source
- If you want to compile the report by yourself you need a LaTex Compiler for your OS and an IDE which makes things easier
- If you want to compile, train and play with our Code you need a python working environment. We used Jupyter Notebooks. The requiered packeges you can see in the Notebooks itself.
cd/you_cloned_repo_location jupyter notebook
Just pull the repo, if you wanna change sth you can ask :)
- Tim Löhr - Coding, Report - GitHub Mavengence
- Timo Bohnstedt - Coding, Report - GitHub Bohniti
See also the list of contributors who participated in this project.
Pretty much the BSD license, just don't repackage it and call it your own please! Also if you do make some changes, feel free to make a pull request and help make things more awesome!
The authors would like to thank Dr. Kede from the City University of Hongkong for good supervising of our group.