Plant feature image recognition
Apply machine learning techniques such as convolutional neural networks to recognising and counting or measuring plant features. This should be implemented using an off the shelf machine learning library such a Tensor Flow. A set of example ground truth images can be provided.
Web interface for sensor data
We have a set of temperature/light sensors that were developed in house (see https://github.com/nppc-uk/wireless_sensors). These measure temperature and light every 5 minutes and store it in a database. This needs a better web interface for exploring the data, an existing interface has already been written in R-Shiny but it gets too slow when used with more than a few hours worth of data.
Gravimetrics web interface
We have a custom developed automatic plant weighing/watering system based around the Raspberry Pi. Configuration and running of the system is done through a command line interface or a rudimentary web interface, both written in Python. The web interface needs a major overhaul with usability improvements, extra functionality to design experiments and data viewing/export. Existing code base at https://github.com/NPPC-UK/Gravimetrics.
Refactoring Octave Image Analysis code
Over the last few years we have developed a set of Octave (GNU’s Matlab clone) routines for analysing plant images. This code is functioning but not in a state where others can easily make use of it. We need somebody to refactor it to improve readability, write proper documentation and examples. Prior Matlab/Octave experience will be absolutely essential for this.