This is an iOS prototype to determine regional fish species on images. 🎣 🎣 🎣 🎣
The model was trained with the help of Tensorflow. For this purpose, 10 different species of fish were examined and trained on the MobileNet_v1_1.0_224 model. Altogether 2000 different images could be compiled, which were evenly distributed on the 10 fish species.
The aim was to find fish pairings, which have certain similarities due to their color, shape and fin combinations. With the help of a confusion matrix, it should be found out how much, and especially which types of fish are wrongly predicted, in order to gain further insights.
This model can differentiate between 10 different fish species:
Download the App by cloning the repository or download the zip file.
Open up the terminal and go to the root of the folder.
cd path/to/bei_fisch_frag_chrisch_version1.6
Make sure that you have Homebrew and Carthage installed.
/usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"
brew install carthage
carthage update
If it was successful, it should look like this:
It can happen, that Xcode is complaining about finding neccessary libraries. Therefore go to the Carthage folder and add CropViewController.framework and PaperOnboarding.framework to Linked Frameworks and Libraries via drag and drop. This should look like this in the end:
This prottotype was made with the help of 2 libraries: