For my senior design project, I am working with two other Purdue computer engineers to build an Android app which can identify the type of bark it is show through the phone camera. This will be done with Koltin and TensorFlow.
To make the app compatible with trees in the Purdue area, we will make custom image dataset for training. We will go to forests in the area with someone trained in tree identification, and build a significant image database for each type of tree we want to identify. This will involve taking a video looking up and down the tree while walking around it. The videos can then be divided up in to images and scaled to the size needed for the neural network to train on.
