Following project constitutes a part of learning process towards completion of the author’s Master Thesis. The main aim was to create a binary classifier that is able to distinguish between triangles and circles with usage of Caffe framework. The repository holds all scripts used during development process with appropriate set of descriptions. Unfortunately, the train network could not be uploaded due to GitHub’s 100mb file size restriction.
Webpage with results:
https://dybiszb.github.io/CNNBinaryClassifier/
- cnn - Holds script for classification process and required .prototxt sturcture. Since taught network is too big for GitHub repository it must be recreated with usage of scripts.
- docs' - Simple data for the results webpage listed at the top.
- imageNetURLs - List of URLs from ImageNet.
- tools - Set of scripts that enables one to recreate the author's development process.
- training - Description of network's structure.
Please refer to directories individual Readme.md for further information.